Printer Friendly

New measures of British service outputs.


1. The growth of service activities

Service sector economics now constitutes a major branch of economic studies although it is a field which no more than a decade ago was variously referred to as the 'poor relation' and the 'Cinderella of academics and politicians alike'. The explanation of the interest recently evinced in the economics of the service sector is to be found in the extent to which service industries have expanded relative to other economic activities. Whilst the tendency for the tertiary sector to grow in comparison to primary and secondary activities has been identified and commented on for many decades special factors, such as the potential for services to generate new jobs, have helped thrust the sector to the forefront of economic analysis in the last few years.

The relative growth of services in the British economy since 1971, quantified in table 1 on the basis of official statistics, could well stand as on outline of events in most developed economies during this period. Between 1971 and 1986 the output of the service sector (defined as distribution, catering, transport, banking and finance, and other services) increased by 2.5 per cent a year, or more than twice the rate, 1.1 per cent, achieved by other economic activities (agriculture, energy, manufacturing and construction). Within the sector commercial services--distribution, catering, financial, business, recreational and personal services--expanded faster than other services, 3.1 per cent compared with 1.7 per cent a year, whilst specific commercial services grew very rapidly indeed--banking and business services by as much as 5.1 per cent per annum. As a result, by 1985, service activities in Britain accounted for three-fifths of GDP and commercial services themselves were responsible for a third of total output.

The importance of service activities is no less when judged by labour force size. On the basis of both numbers of employees, and by numbers of employees plus the self-employed, by 1985 services accounted for 66 per cent of Britain's labour force compared with 53 per cent in 1971. Commercial services alone provided jobs for 37 per cent of all employees in 1985.

2. The need for new measures

At a time when the bases, compilation, accuracy and usefulness of a range of official economic statistics have come under close scrutiny, the measurement of service outputs has not escaped attention. Recently there has been some shift away from a simple acknowledgement that service output measures may be subject to a degree of unreliability towards an apprehension that any errors they contain might lead to a downward bias in the measure of service output, and therefore, GDP growth rates. It has in fact been suggested that output increases in services may have been underestimated in recent years by as much as 2-1/2 per cent a year implying that annual growth of GDP itself should have been substantially higher.

For both conceptual and practical reasons the outputs of many service activities are notoriously difficult to quantify both absolutely and in terms of change over time. Yet because of their size and growth it is especially important that a reasonable degree of accuracy must attach to service output measures which should reflect current best practice given the data and resources available. If this is not the case then, because of the weight of service activities in the total economy, the accuracy of the registred change in GDP and national productivity, the relative contribution to growth of service and non-service activities, the pattern of structural change within the service sector and the policies to which these various phenomena have given rise, must all be called in question.

This article assembles some of the principal results which have emerged from a study of alternative output measures for British service industries. The study reviews the methods currently used by the CSO to measure real output changes in specified services and seeks to devise, develop and implement new measures for these activities. For this purpose an empirical approach has been adopted. Whilst basic conceptual considerations are taken fully into account in that the alternative measures which are compiled can be integrated into the national accounting framework, it is felt that the theoretical complexities associated with service output measurement have received due attention elsewhere whilst very little has been done at the practical level.

When compiling new service output measures the two principal uses for which they are employed must be borne in mind. The original raison d'etre for measuring service sector output is for use, in conjunction with output indicators for industrial activities, to yield a measure of real growth in total GDP. This is the primary role for which the CSO devises and compiles output measures for service sector activities. Increasingly, however, as the weight of services in the economy has grown service output indicators are now used extensively, in conjunction with labour force and other relevant economic series, for the analysis of long-term developments in individual service industries. Their employment in the latter context accentuates the requirement that they fulfil minimum reliability criteria: it is possible that offsetting errors in individual service industry output measures may modulate their impact on the reliability of overall GDP measurement but this does not apply when attention is focused on the analysis of developments in a specific service industry.

The results obtained in the present exercise suggest that official measures of GDP growth may well have been marginally understated as a result of the methods used to track service output changes. Even more significantly, perhaps, the alternative measures which have been compiled portray patterns of development for some individual commercial service industries which are very different from those yielded by official output measures.

When considering alternative output measures attention has been directed first at the strengths and weaknesses of official practice as a prelude to an attempt to improve the rationale and reliability of the resulting output indicators. However in some cases it is necessary to regard the new measure simply as an alternative to the official index: an alternative with a different, rather than superior, base and/or derived from quite separate data sources. Another feature of the search for alternative indicators is that it has been conducted with an eye to the improved reliability of service output measurement in the future. To some extent this means that data series which are available only for recent years have been drawn upon. Nevertheless a major objective of the study has been to assess the extent to which the official picture of past service sector developments is changed if alternative output measures are used. The alternative measures have therefore been carried backwards in some cases to 1971, in others to 1973 and for a large number to 1978, using 1985 as the base year. There are three reasons for the latter choice. First data limitations mean that all series cannot be uniformly taken back to a common early date so that a recent year must be adopted as the base. Secondly at the time when the research was initiated 1985 was the latest year for which data--whether values, 'quantities' or 'weights'--were generally available. Thirdly, the CSO was then in the process of re-basing its output indicators using 1985 GDP weights.

The new output measures draw much more extensively on unofficial data sources than do those compiled by the CSO. Broadly such sources can be divided into two types: those which yield data series covering an industry-wide set of activites; and those which contain data relating to individual service industry organisations and firms. Typically the assistance of the latter was enlisted to provide weights with which industry-wide quantity series for different kinds of output can be combined to produce a single output measure for the service sector in question.

It is to the credit of the CSO that there has been a continuous attempt to modify service output measures in the light of methodological developments and the emergence of alternative data sources. Whilst this has had the desired result of improving the reliability of service output measurement inevitably it also introduces inconsistencies and breaks in the time series where--either on practical or other grounds--the new methods have not been carried back to earlier years. A feature of the new measures postulated below is that every effort has been made to ensure temporal consistency in the series which they yield.

It has not proved possible to devote to 'quality' aspects of service output the attention they properly merit. In part this reflects the fact that the incorporation of quality changes in output measures may be even more difficult in services than in goods-producing industries where frequently its assessment is equally neglected. This in turn is due to the consideration that the quality of a given service varies greatly (no two football matches are the same in this respect) and in many cases (as with say opera performances) is highly subjective. Nevertheless the results do have implications for this aspect of service output measurement, especially in the case of the catering trades. Were quality aspects taken into account more fully, further consideration would need to be given to the relative merits of measures based on deflated value series and service output units.

3. Coverage of the measures

The search for alternative output measures has been restricted to the commercial service industries, essentially distribution, catering, financial, business and personal services, for which sector table 1 suggests that output has risen especially rapidly. Within this group of commercial services the CSO distinguishes for output measurement purposes as many as 37 service industries, output changes in which are currently assessed by a total of 114 indicators. These services accounted for some 30 per cent of total GDP in 1985 and about a third of people employed in Great Britain.

Given the vast range of service activities which must be covered by official output measures it is natural that the CSO makes use of readily available, relevant, series to yield the required indicators, series which themselves are normally official in the sense that they are compiled by other government departments. Since the quest for new measures has entailed a search for and detailed examination of, alternative data, involving the identification, location and perusal of non-official sources and also the co-operation of individual firms and organisations it proved necessary to regard the full range of commercial service activities as no more than a frame from which individual service industries could be selected for intensive analysis.

When choosing industries from this list for which alternative measures of output change are identified and implemented, several criteria have been applied. A principal consideration has been to direct efforts to services where, on the face of it, current practice appears to be weak. At one end of the scale this principle tends to divert attention away from activities such as retailing, where the measures are generally regarded as being relatively sound, towards such services as advertising where the official output indicator is based on an employment series. Secondly some priority was given to activities for which at first blush--sometimes deceptive--alternative measures appeared feasible. Thirdly an attempt has been made to cover the larger service industries such as banking and insurance, and ceteris paribus, attention was focused on service activities which, a priori, are thought to be rapidly expanding, a criterion which points to the inclusion of, especially, financial and business services.

Since a systematic attempt has been made to base alternative output measures on numbers of service units produced, such service industries as estate agents, stockbroking and legal services where official output measurement is already founded on numbers of service units--of property transfers, transactions, and court proceedings--have not been considered as candidates for alternative measures.

The service industries for which new measures were ultimately compiled fall neatly into three groups. First, there are ten selected financial and business services: banking, building societies, finance leasing, hire-purchase, insurance, accountants, architects, advertising, computer services and construction plant hire. Secondly eight recreational services have been covered: broadcasting, theatres, libraries, museums, professional sports, participatory sports, local authority leisure centres and betting and gaming. Thirdly, the five catering trades--restaurants, public houses, clubs, catering contractors and hotels--have also been included.

4. Alternative output measures: numbers and values

Continuous CSO action to modify and improve service output measures means that they have developed from year to year in an essentially ad hoc manner with considerable diversity in their conceptual underpinning. Output measures for the commercial service sector rely heavily on deflated value and employment indicators and whilst output measurement for some service activities, noted above, is based on counting numbers of service units, such practice applies to only a fraction of the output of the commercial service sector. In fact it has emerged that the use of this type of measure can be extended to a considerable number of commercial services and an attempt has therefore been made to generalise and implement this approach, especially in the case of the financial and recreational services. This method can be summarised as the compilation of [sigma]poq1/[sigma]poqo where q represents the industry's output quantities (in the present context, service output units) p the industry's output prices and 0 and 1 refer to the years compared. This is 'the traditional [measure] used for multi-product industries wherein an index of production is constructed from a weighted sum of various outputs produced by the industry'. An indirect way of arriving at the same result, that is measuring [sigma]poq1/[sigma]poqo, is to calculate the following quotient: [sigma]p1q1 / [sigma]poqo / [sigma]p1q1 / [sigma]poq1 = [sigma]poq1 / [sigma]poqo where [sigma]p1q1/[sigma]poqi is the deflator. This is the basis of deflated value measures of output change, frequently used for service industries, where the indicator is derived by deflating a turnover series by the relevant price index. Such a deflated value approach to output measurement will yield the required result--as measured by the traditional direct method based on weighted quantities--only if the deflator, the price index, relates specifically to the products or services in question. This condition is not fulfilled in the case of many service output measures where for practical reasons some form of general price index is used. It is partly to remedy this weakness in service output measurement that, wherever possible, alternative measures have been based in this exercise on the traditional direct methodology using numbers of service output units.

A fundamental requirement for the use of this 'traditional' or 'direct' approach is that, for purposes of practical measurement, the outputs of the industry or service in question should be relatively homogeneous. No economic activity yields a single, identical, product or service so that in practice it is sufficient if the bulk of an industry's activity is represented by a limited range of output types. It is their supposed inability to fulfil this criterion which has discouraged the application of the traditional, direct, measurement methodology to service activities. Generally it has been assumed that there exists, virtually, an infinite variety and range of outputs in the case of most, especially financial, services: that no two life insurances, mortgages, bank accounts, and the output activities associated with them, are the same.

It is argued that this view is based largely on a misconception, associated with what, for want of a better term, we shall call the 'digit illusion', and that in fact many service industries--including betting and gaming--fulfil the basic criterion for the application of the traditional, direct measurement methodology in that the bulk of their output is accounted for by a limited range of essentially homogeneous activities. For the impression of enormous diversity in the output of most financial services derives from the fact that the values associated with a specific activity--the sums assured by a life company, sizes of mortgages provided by a building society, the amounts loaned under hire-purchase agreements by a finance house, the values of cheques cleared by a commercial bank, the car insurance premiums charged by an insurer etc--do have an infinitely large range.

It has been assumed in the search for new output measures that such differences can be essentially ignored. In the case of mortgages, for example, the output associated with writing 500,000 pounds sterling in a deed is, with some relatively minor qualifications, little different from that associated with inscribing 50,000 pounds sterling or even 5,000 pounds sterling; differences in the numbers of digits required in this and other financial instruments--cheques, insurance policies, hire-purchase agreements, betting slips--have relatively little bearing on the amount of output involved.

In brief for a given type of financial activity it is normally legitimate, when measuring output, to ignore the 'digit illusion'. This has the effect of rendering homogeneous, by the above definition, a wide range of financial activities, making them amenable, in principle, to the traditional direct, measurement methodology. It means, by way of illustration, that output changes for the life insurance industry can be based on changes in numbers (not values of sums assured, expenses or premiums) of policies. It must be stressed however that a distinction must be drawn, and numbers counted, for each major type of service/activity provided by the service industry in question where value added per service unit varies significantly between activity types. This means, for instance, in the case of the life insurance industry, that numbers of policies must be counted separately for ordinary life insurance, industrial life, annuities and personal pensions.

It also means that attention needs to be drawn to a feature of this approach which is peculiar to financial service industries: the distinction between stocks and flows. Clearly, the resources devoted to, and therefore the amount of output associated with, the annual 'maintenance' of pre-existing mortgage will be substantially less than required when initiating a new mortgage. In principle, and where practical, it is as important therefore to draw a distinction when measuring output between numbers of existing mortgages issued in a given year (the 'flow') and the numbers of existing mortgages (the 'stock') as it is to count separately the numbers of, say, different kinds of policy issued by life insurance companies. Indeed the same distinction needs to be drawn in the case of life insurance--between new policies and policies in force--and in principle at any rate in many other financial service industries: between new and existing contracts for hire-purchase and finance leases for instance.

The practicability of extending the traditional measurement methodology to commercial services thus hinges, in the case of each activity, on the availability of two kinds of data: industry-wide series of numbers of each type of principal (homogenous) product, the series q; and appropriate weights with which these various product series can be combined to yiedl an aggregate output measure for the service industry in question, the series p.

To a degree inadequacies in the measurement of service output changes simply reflect the relative lack of interest with which they were regarded in the past and a consequent failure to devote sufficient resources to the collection of basic output data. Also, it is to some extent, due to the 'insubstantiality' of most services, the fact that they 'pass away' in the moment of production, that a systematic effort has not been made to base their output measurement to a much larger extent on numbers of service units produced. In fact regardless of whether or not a service has any kind of physical embodiment, in the vast majority of cases there exists a physical record or token that a particular service has been performed. The outputs of all kinds of commercial service industries are very well 'documented'--by the ticket required for entrance to Wimbledon, the bill for a meal, the contract which records the finance leasing of a fleet of aircraft--in a form which in principle allows the numbers of each type of service to be recorded. In this sense, and at this lelve, the output of service activities is registered at least as well as the products of industrial pursuits. The nig universal existence, at the 'grass roots' level, of this detailed record of service sector outputs, to an ever increasing extent in computerised form, augurs well for the future extension and development of the kind of direct output measurement advocated here. This comprehensive data base has not yet been properly exploited to yield aggregate series of the numbers of each type of service produced on an industry-wide basis. However research and enquiry have revealed that representative bodies of various kinds at the industry level--especially trade associations--have come to regard the compilation of aggregate numbers of each kind of service produced, using these data bases, as one of their primary tasks, and such sources have been widely drawn upon.

That they may not be based on series for numbers of service units which are fully comprehensive for the industry in question should not be regarded as a major weakness of the alternative output measures which are presented in the following section. Usually only the major 'products' are counted and in this respect the circumstances are no different from those which exist when the method is applied to industrial activities. For both industrial and service activities so measured the implicit assumption is the same: that real changes in the non-covered outputs parallel the aggregate change measured for the covered operations.

Unfortunately, with few exceptions, service industry organisations, do not collect, process or publish, the kind of data (relating to expenses for example), which are needed to obtain the 'weights' required for aggregating these various output series into a single measure for the industry in question. In these circumstances it has proved necessary to have recourse inter alia to the goodwill of individual (if major) firms in order to derive suitable weights which must be accepted as typical for the service industry in which they operate.

Wherever possible alternative measures have been derived from an application, along the above lines, of the traditional, direct methodology based on numbers of service output units. However even when allowance is made for the 'digit illusion' some service industries--such as architecture and accountancy--remain in that class where output units are, in essence, infinitely variable in nature. In these cases an attempt has nevertheless been made to upgrade the output measure, usually by identifying and compiling a more appropriate price deflator.

5. The results

This section presents the results of the search for new output measures for the selected commercial activities. Initially the new output series are presented separately for the three major service complexes--financial and business services, recreational services and the catering trades--which have been covered and compared with the official measures. In each case the nature of both the alternative and official series are stated only in brief: it is planned to describe the methods used in full detail elsewhere. The section concludes with a consideration of the implications of the results for future developments in the measurement of service output.

Financial and business services

In 1985 the ten financial and business services for which new output measures have been compiled contributed, in total, about 35 pound sterling billion to GDP and employed approximately a million people. In the case of banking, the largest of the services in this sector, it proved possible to base the new measure wholly on numbers of service units--numbers of accounts, clearings, cahs and credit card transactions. This contrasts with official practice in which indicators derived from deflated series for deposits and loans, as well as employment, carry the bulk of the weight. In life insurance, also, numbers of policies of various types, distinguishing between new policies and policies in force, replace the official series comprising deflated consumer expenditure. Similarly the new building society measure is founded wholly on numbers of service units--shareholders, existing borrowers and new loans--instead of, as is the case with the official measure, a combination of deflated liabilities, employment and numbers of advances. For general insurance, a measure reflecting broadly the stock of insurable assets has been substituted for the official indicator based on deflated premiums.

In the case of two services, architects and advertising, a deflated value series places officla measures based on employment. The primary feature of the new measures used for accountants, computer services and hire-purchase, is the use of deflators which have been specially constructed to reflect more closely than those officially employed developments in the activities in question. The official practice for both construction plant hire and finance leasing has been to base indicators on output developments in user industries. Instead of this approach the new measures are derived from gross output in the case of construction plant hire and the value of assets newly leased and of the stock of leased assets in the case of finance leasing, each deflated by price indices specifically compiled for the activities in question.

The new measures for the covered financial and business services are set alongside the official indicators in table 2. The basic date series from which these were derived are such that not all of them cover the full span of years, 1971 to 1986, so that it is convenient to combine them into three groups: five industries--building societies, hire-purchase, insurance, advertising and construction plant hire--for which growth rates are available for the whole period 1971-86; six industries--the above five together with banking--where the information is shown for 1973-86; and all ten covered financial services--the above six plus accountants, architects, finance leasing and computer services--for which alternative and official growth rates can be compared for the period 1978-86.

In the case of the first group, for which measures extend over the full period 1971-86, whilst there are some very significant differences between alternative and official growth rates--for building societies the former is appreciably higher than the latter whilst the opposit is true for hire-purchase--the (unweighted) arithmetic average growth rate for all five services, at 4.1 per cent per annum, is identical for both alternative and official measures. The same generalisation applies to the group of six industries over the period 1973-86: for hire-purchase, advertising, construction plant hire and banking the two measures yield quite different results but these contrasts offset each other so that the (unweighted) average alternative growth rate is identical with the average official growth rate.

Results for the group of ten industries, relating to the period 1978-86, paint a very similar picture; of individual industry diversity but overall similarity. In this case only for building societies, insurance, advertising and accountants do alternative and official methods yield rates of output change that are at all akin. Yet, overall, the (unweighted) average alternative growth rate, 7.4 per cent per annum, is not much greater than the 7.0 per cent registered officially. The official measure for finance leasing is constructed to minimise distortion in GDP measurement whilst the alternative measure has been compiled in such a way as to enhance the reliability of the industry's growth indicator per se. If this service industry is left out of account, the average alternative growth rate, at 6.6 per cent, is below the average official rate, 7.6 per cent.

This set of results contains a further point of interest. Whilst, as pointed out, there are substantial differences in the case of most industries between the alternative and official measures, the inter-industry growth patterns are remarkably similar whether gauged by alternative or official indicators: again leaving out finance leasing the correlation coefficient for the remaining nine pairs of, alternative and official, growth rates is +0.95.

Although comparisons of unweighted average growth rates can be quite instructive more attention should be focused on differences between alternative and official measures revealed by weighted aggregate indices of the kind shown in table 3. These output indices, derived from the alternative and official measures for individual industries and relating to the same groups of industries and periods specified in table 2 have been obtained from an aggregation procedure which is based on 1985 contributions to GDP. The results contained in this table are summarised in the form of average annual rates of growth in table 4 which also contains growth rates obtained from weights derived from 1980 industry contributions to GDP and 1985 industry employment levels, to test the sensitivity of the results to alternative weighting systems.

In the case of the five industries for which there is full coverage over the period 1971 to 1986 the alternative growth rate is 4.4 per cent and the official one 4.8 per cent on the basis of 1985 GDP weights. A discrepancy of no more than a tenth between the two versions is also suggested by 1980 GDP weights. This group of five industries accounted for only 20 per cent of the contribution to GDP and an estimated 27 per cent of employment of all financial and business services in 1985. In contrast the group of six were responsible for 56 per cent of output and an estimated 42 per cent of employment and the group of ten for as much as 74 per cent of output and 56 per cent of employment. Greater significance should therefore be attached to the results obtained for these latter two groups.

All three weighting systems show that for the group of six industries, over the years 1973 to 1986, the official measure understated growth by about a fifth when compared with the aggregate alternative measure; much the same picture emerges for the group of ten industries during the period 1978 to 1986 where, on the basis of both 1985 and 1980 GDP weights, the official measure, compared with the alternative, underestimates growth by about a fifth.

When considering this disparity two industries merit special attention: banking which is by some way the largest of the ten industries; and finance leasing on account of the fundamental difference, referred to above, in the methods underlying the alternative and official measures. In fact if banking--with alternative and official growth rates of respectively 8.2 and 6.5 per cent per annum for the period 1978 to 1986--is left out of account the picture hardly changes at all: the alternative aggregate growth rate for the remaining nine industries over the period 1978 to 1986 (based on 1985 GDP weights) emerges as 8.0 per cent per annum compared with an official growth rate of 6.0 per cent. This is in sharp contrast to the effect of excluding finance leasing. In this case the alternative growth measure for the nine remaining industries falls to 7.5 per cent whilst the official measure rises to 7.2 per cent so that most of the differential disappears.

Recreational services

In 1985, recreational services, SIC class 97, contributed almost 6 billion pounds sterling to GDP and provided employment for 430,000 people. Apart from cinemas, for which the official output indicator is based on numbers of attendances, the measures used by the CSO rely for the most part on deflated series of turnover or consumers' expenditure on the service in question. In the case of local authority libraries, however, employment-based indicators are used.

In the search for alternative measures it has proved possible, for these services, to base the new indicators largely on numbers of service units. Thus in table 5 the output index for 'group' 974, broadcasting and theatres, reflects changes in numbers of radio and television hours transmitted (in the absence of adequate series for numbers of viewing/listening hours) and of attendances at theatres etc, whilst that for libraries and museums has been founded on numbers of books issued and numbers of visitors.

The new measures for sport, 'group' 979, an activity that accounts for almost half of the output of recreational services, has been derived from a variety of indicators which reflect, inter alia, changes in numbers of attendances at paying spectator sports, sports hours transmitted by television and radio, membership of participatory sports clubs and measures which trace changes in numbers of betting slips and football coupons. Only in the case of local authority leisure facilities was recourse made to conventional deflated value series.

The output series set out for cinemas and for authors and artists in table 5 are those used officially: that for cinemas being already based on numbers of service units whilst no alternative measure could be readily identified for 'group' 976.

For comparison with the overall official measure for recreational services the five component output measures in table 5 have been combined into a single output index using weights based on 1985 contributions to GDP.

The official measure suggests that the output of this service industry complex grew, between 1973 and 1988, at an average annual rate of 2.9 per cent; the alternative points to a significantly slower rate of increase of 2.0 per cent. One factor which may go some way towards explaining this discrepancy is that whilst in principle the alternative measure includes local authority libraries and leisure facilities throughout the period, the official indicator embraces these activities only since their transfer to class 97 of the 1980 SIC. Yet it can be seen from table 5 that even in recent years the official measure has registered a significantly higher rate of growth for this industry than that shown by the alternative measure.

The catering trades

In 1985 the catering trades' contribution to GDP totalled 7-1/2 pounds sterling billion and the industry employed more than a million people. Output measures for these trades are summarised, in terms of average annual growth rates, in table 6. In the early years of the period covered the official indicator was based on turnover data supplied by a voluntary panel of catering organisations deflated by price indices specific to each of the catering trades. From the beginning of the 1980s the official measure has been derived from changes in the margins of the individual catering trades as recorded in the DTI's annual catering trades' enquiry, again deflated by specific price indices.

Although in principle it should be possible to measure output in the catering trades on the basis of numbers of service units--hotel bed occupancy, numbers of bar transactions etc--attempts to use this approach proved, for practical reasons, to be fruitless. Therefore to test the sensitivity of the official measure alternative output series were constructed using the price indices officially compiled but applying them throughout to turnover data (as opposed to the gross margins used officially in recent years) yielded by the catering trades inquiries, interpolating results for early years where no such inquiries were conducted.

It is clear from table 6 that these alternative measures alter the picture quite significantly for some trades. Whilst for restaurants, clubs and catering contractors changes in output are not greatly different whether measured on the official or alternative bases, sharply contrasting developments emerge for public houses and the hotel trade: the substantial growth recorded for public houses by the official measure virtually disappears with the alternative, whilst the opposite occurs in the case of hotels. These quite significant modifications to the subsectoral picture, in effect, cancel out: on both official and alternative methods catering output increased over this period by about 1 per cent a year.

The reliability of catering output changes cannot, however, be allowed to rest there. For when these changes are compared with labour force developments there would seem to have been substantial long-term declines in labour productivity: taking the official output measure in conjunction with changes in the number of employees in catering, labour productivity appears to have fallen by as much as 1.7 per cent a year between 1971 and 1986.

To help determine whether these results reflect real developments in this service industry or whether they arose from statistical series which are sufficiently unreliable as to produce a misleading picture of events, alternative, more sophisticated, full-time equivalent labour force series have been constructed. Since both part-time working and self-employment are important features of the catering trades' labour force, allowance for developments in both these aspects has been made in the FTE series shown in table 6.

Whilst the introduction of FTE labour force series improves the productivity picture there nevertheless remains an apparently substantial fall in catering productivity: an annual average decline of 1.7 per cent a year using numbers of employees is modified to a deterioration of 1.2 per cent on an FTE basis. Of the individual trades only catering contractors emerge with enhanced FTE productivity, displaying an improvement of 1.7 per cent a year. The other four trades register deteriorations: restaurants by 2.1 per cent a year, public houses 0.1 per cent, clubs 3.6 per cent and hotels 1.2 per cent.

It is theoretically feasible, on the basis of most formulations of the production function, that in certain circumstances output in any economic activity might increase over the long term at a rate below that registered for labour inputs. This could arise in particular if there were a fall in capital intensity but such a development does not appear to have been a feature of catering during this period. A measure compiled for the industry suggests that between 1971 and 1986 the stock of equipment grew at an annual average rate of 3.6 per cent a year which, in conjunction with an annual rise of 2.3 per cent in the FTE labour force, points to a 1.4 per cent annual increase in capital intensity.

In these circumstances it is difficult to accept that either the official or alternative output measures presented in table 6 adequately reflect output growth in the catering trades. A probable explanation of this deficiency is that the measures fail to take sufficient account of developments, in some general sense, in the quality of catering output, be it a reflection of changes in the range of services offered, the milieu in which they are provided, the nature of the food and drink supplied or the standards of customer service.

It is relevant that following comparatively little change in its size during the 1950s and 1960s the catering industry's labour force has grown substantially over the last two decades, especially during the early 1970s. These developments have generally been explained by contrasts in the nature of the labour market between, on the one hand, the earlier decades and on the other the 1970s and 1980s. The earlier prevalence of full employment meant that because catering work was notoriously unattractive and ill-paid the industry was able to adjust its labour force towards its basic manning requirements. This kind of effect was reinforced by the impact of the selective employment tax which led to a significant fall in catering employment when it was implemented during the later years of the 1960s and a rapid increase in the early years of the 1970s after its abandonment.

This thesis is also supported by case studies of the impact on the industry of technological innovations (computers, microwave ovens etc). A recent survey of organisations operating in the catering trades revealed that rather than causing redundancies 'the time saved by new technology and improved productivity was being used wherever possible to enhance customer service'. A stagnant productivity performance in catering is equally difficult to reconcile with the rapid growth of both take-away food facilities and self-catering accommodation.

In the event it has been decided to use labour inputs, to be precise the FTE measure shown in table 6, as a surrogate output measure for this industry. It is felt that whilst far from ideal it provides a less unsatisfactory output measure, over the period in question, than either the official index or the alternative indicator compiled above. On this basis an annual average growth rate of 2.3 per cent emerges for catering output, hardly an unreasonable achievement for what is largely a leisure industry during a period when GDP increased, on average, by 1.7 per cent a year. Furthermore when applied to the individual catering trades this labour force based method yields results, for changes in real output over these years, which are intuitively much more plausible than those derived from either the official or alternative methods. It suggests that between 1971 and 1986 output in all five trades increased substantially: in restaurants by 2.4 per cent a year, in public houses by 2.5 per cent, in clubs by 3.1 per cent, in catering contractors by 2.8 per cent and in hotels by 1.6 per cent.

6. A new output measure for all covered


The alternative output measures compiled for financial services, recreational services and the catering trades are combined and compared with their official counterparts in table 7. Group 1, for which the series relate to the full period 1971-86, comprises five financial services (building societies, hire-purchase, insurance, advertising and construction plant hire) and the catering trades. Group 2 covering the period 1973-86 includes in addition banking and recreational services. Whilst group 3, for which the output indicators are available for the period 1978-86, embraces all the service industries for which alternative measures have been compiled. These aggregate output indices, both official and alternative, are derived from those presented in the relevant tables using as weights the 1985 contribution of each activity to GDP. The results are summarised in table 8 in the form of average annual rates of change. To check the sensitivity of these final results alternative and official measures of growth are also compared using as weights the services' 1985 employment levels and 1980 contributions to GDP.

The picture which emerges is one in which the alternative measure consistently points towards a higher growth rate than that indicated by its corresponding official index. Moreover the size of the differential varies comparatively little regardless of the group, period or weighting system to which attention is directed. If most attention is attached--in view of the size of their coverage--to the results for groups 2 and 3, it appears that compared with the alternative methods the official indicators have understated growth by between 13 and 18 per cent. Given that in most cases the alternative measurement methods differ radically from those officially employed it is perhaps remarkable that, overall, the discrepancy should be no greater than this.

To what extent, as it stands, would the official measure of GDP growth be changed if the alternative measure is substituted for the official indicator? The answer is: very little. Basing this adjustment on the alternative/official differential revealed for group 2 the annual average GDP growth rate for 1973-86 is unchanged at 1.3 per cent per annum, and using the group 3 results the annual growth rate for 1978-86 rises from 1.6 per cent to 1.7 per cent. If the disparity between official and alternative growth rates depicted in table 8 held for the commercial services excluded from this exercise then the official GDP growth rate of 1.3 per cent registered for the period 1973-86 is boosted to 1.5 per cent per annum and the GDP growth rate of 1.6 per cent for the years 1978-86 is raised to 1.9 per cent a year.

There are good reasons for supposing that these upward revisions to the GDP growth rate will understate rather than overstate the actual shortfall. In the first place output measures which lean towards numbers of service units will to some extent be biased downwards compared with those derived from deflated sales or turnover values because the data sources in question tend to cover any new services in the latter but not the former. Secondly whilst official output measures which rely on employment series usually incorporate some productivity improvement, this has not been applied to the FTE based measure compiled above for catering so that it may well underestimate actual output growth in this activity. Thirdly salary indices which have been used as surrogate price deflators in the case of accountancy and computer services will tend to overestimate actual output price increases--and therefore bias downwards the resulting output measure--since no allowance is made for the impact on unit labour costs of any productivity improvements in these services.

While the implications for GDP measurement of the substitution of alternative aggregate service output indicators may not be unduly significant it is clear that they are more substantial for the commercial service sector as a whole. They are even more significant for individual service complexes. Although the new measure for financial and business services differs relatively little from the official measure, especially if account is taken of the contrasting conceptual treatment of finance leasing in the respective approaches, the alternative obtained for recreational services points to a significantly lower growth rate and that compiled for catering is about twice the official growth rate. Contrasts between alternative and official measures are even more pronounced for individual service industries.

7. Future developments in service output


Substitution of the new output measures for those officially compiled would have made little difference to GDP growth rates in recent years. In effect discrepancies which emerged between alternative and official output measures for individual service industries have tended to offset each other. There is no guarantee that this condition would hold in the future and for this reason alone the search for alternative, improved, measures of service output change should be pursued.

In contrast significantly large differences between alternative and official output measures have emerged for particular services and these have very important implications for the analysis of economic developments in individual service industries. The validity of the economic analyses to which these activities are now widely subjected depends crucially on the accuracy of the output and productivity measures that are employed to map their development over time. For this reason, especially, new and improved service output measures are required.

A lesson of the research which lies behind the above results is that new service output measures can be further developed both extensively and intensively. Whilst the service industries examined were chosen as systematically as possible on the basis of specific criteria, only limitations imposed by time and resources prevented other service activities being subjected to a similar kind of scrutiny. Moreover in many of those service industries which have been surveyed scope for further measurement improvements remain. In this connection attention needs to be drawn to the results obtained for building societies and the catering trades. In the case of the former serious doubts attach to the alternative measure which, although it yields a growth rate higher than the official one, nevertheless implies, as it stands, a long-term fall in labour productivity in this service industry. Also it is clear that the search for improved catering output measures needs to be continued: the labour force indicator adopted for the purpose of this exercise must be regarded as no more than a stop-gap strategem though on the basis of existing statistical sources there is at present little immediate prospect of new measures being based on numbers of service output units. None the less the data deficiencies are essentially practical, not conceptual, in nature and with appropriate statistical arrangements could no doubt be remedied eventually: already there is some experience of collecting data relating to hotel room utilisation which some time in the future might form the basis of a service unit-based measure.

Generally there seems to exist a large reservoir of data relating to both output indicators and weights, collected by a variety of service industry organisations, which lies outside the purview of official sources. It is difficult to avoid the impression that the current exercise has barely scratched the surface of this fund of material and that, in particular, much is available at the firm level which has yet to be systematically collated in a way that it can be used to measure developments for whole industries. Certainly it is clear that within the firms in question most service outputs are documented in detail, increasingly in computerised records, so that alternative measures based on numbers of output units should become progressively more practicable.

There appear to be two basic options for tapping into this kind of data base. One would comprise an official periodic survey of each of the main service industries in which, as in the case of some industrial censuses, questionnaires would seek information about numbers and values of those output units which are best measured for each service. The inquiries into the distributive, catering and service trades periodically conducted by the Business Statistics Office might be developed in this direction. The alternative is to adopt the approach followed, in the main, in this exercise and base new output measures on the information which is collected by trade associations and other industry organisations. This option is less attractive than the systematic collection of the requisite data, on a comprehensive and consistent basis, by means of an official census.
COPYRIGHT 1989 National Institute of Economic and Social Research
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Smith, A.D.
Publication:National Institute Economic Review
Date:May 1, 1989
Previous Article:Two nations of shopkeepers: training for retailing in France and Britain.
Next Article:Obtaining estimates for the standard errors of long-run parameters.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters