Manufacturing productivity levels in France and the United Kingdom.

Van Ark, Bart

1. Introduction

International comparisons of levels of labour productivity are rare in the field of productivity analysis. in the case of Anglo-French comparisons, for example, it has already been widely established that the French economy was more slowly transformed from an agricultural economy into an industrial society than the United Kingdom; and that since the last world war manufacturing output has increased much faster in France than in Britain. The aim of the present study is to complement previous comparisons of growth rates of manufacturing productivity in Britain and France with estimates of the current differences in the levels of output per person-hour worked in a dozen branches which constitute the manufacturing sector. This study adopts an original approach to estimating relative productivity levels between France and the United Kingdom. Firstly, the estimates are based on returns by firms on output and inputs made for the census of production and for official industrial sample surveys in each country. Previous comparisons of productivity levels in Britain and France relied on secondary and more aggregated sources for output and inputs, for example, the national accounts and employment statistics. Secondly, the output estimates in this article are converted into the same currency using ratios of ex-factory prices-or unit values-for manufactured products. Most other studies have used either currency exchange rates, or at best purchasing power parities, which relate to retail prices and prices of capital goods. (2)

Despite such advantages, the estimates in this article still have to be treated with some caution in view of the decentralised way in which inquiries are carried out for the Enquetes de Branches, the main source of French product information for this article. As a result the percentage coverage of product sales which could be matched between the two countries to derive unit value ratios was limited. Nevertheless, the estimates in Section 2 of this article provide clear evidence of a substantial manufacturing productivity gap between France and Britain though smaller than that observed in previous studies using the alternative methods and sources mentioned above.

Before presenting the results of this study in detail, some background information on the two countries needs to be given. At present the size of the British and French population is very similar at around 56 million, but France has an area over twice that of Britain. For 1989 average per capita real income in France has been estimated at about 2 per cent above that in the United Kingdom for 1988, down from 1 0 per cent above the British level in 1980. (3)

Relatively higher per capita income levels in France were achieved only during the recent postwar period. Up to around 1950 the French economy remained dominated by a large and inefficient agricultural sector. (4) Although it would be wrong to claim the absence of any form of industrial development in France, it seems clear that 19th century French industrialists were slow in adopting mass production techniques in their factories, which were so characteristic of the early ages of industrialisation in Britain. (5) French per capita income was no more than two-thirds to three-quarters of the British level between 1870 and 1950 (Maddison 1987). Rough estimates suggest that productivity levels in both agriculture and industry were also much below those in Britain.(6)

The economic growth patterns in France and the United Kingdom since the last world war are also often contrasted. Between 1950 and 1973, total real gross domestic product in France increased much faster at 5.1 per cent a year on average compared to only 3.0 per cent in the United Kingdom (Maddison 1989). As the estimates in Section 3 of this article show, the growth performance of the manufacturing sector in France was so impressive that the productivity level in this sector was well above that in the United Kingdom by about 1973.

It has been suggested that the better economic performance in France was partly reached because of more stable macroeconomic policies, which were of a more dirigiste nature compared to Britain. (7) On the other hand, industrial and trade policies in the two countries were similar in the two countries, though the emphasis was sometimes different. Governments in both countries supported strategically crucial industries of a high-technology nature, such as aerospace, nuclear engineering and electronics; they increased subsidies to ailing industries such as steel and shipbuilding; and at times they both heavily intervened in the external sector to promote exports and restrict imports of competing goods.

This paper concentrates on relative productivity levels in French and British manufacturing during the last two decades. Both countries suffered heavily from the oil crises of the 1970s, and faced similar structural adjustment problems in the early 1980s. A previous Institute study on economic policy in France and Britain concluded that France shares with us [Britain] to a greater or lesser extent the handicaps of an outdated industrial structure, an inflexible market system and deep political divisions' (Barker, Britton and Major 1984, p. 81). it is therefore to be expected that the productivity differential between France and the United Kingdom estimated in this study is smaller than that shown in previous comparisons by the National institute between the United Kingdom and Germany (Smith, Hitchens and Davies 1982) and, more recently, between British and Dutch manufacturing (Van Ark 1990). The estimates in this article are based on the same procedures as those applied in the latest study to ensure their comparability.

2. The productivity gap between British and French manufacturing: methods and sources In this study manufacturing labour productivity in France and the United Kingdom is estimated on the basis of information obtained from each country's census of production for 1984. Table 1 shows the results for the comparisons of productivity for 13 manufacturing branches. For each country, output is expressed in its own currency and in the currency of the other country. The conversion from one currency to the other was made on the basis of average unit value ratios (UVRS) for ex-factory product sales. As some manufactured products are more strongly represented in France and others more strongly in the United Kingdom, the results depend on which country's quantity weights are applied to obtain the average UVR. In general such differences are quite small, so that one can safely work with a (geometric) average of French and British weights, as presented in the last column of table 1. The estimates are adjusted for differences in the number of annual hours worked per employee, which were 9 per cent higher in British manufacturing compared to France (see annex A).

In manufacturing as a whole output per person-hour worked in France was estimated at 31 per cent above the level in the United Kingdom in 1984. Before discussing the results in more detail, the accuracy of this estimate needs to be considered more carefully. There are two major difficulties in cross-country comparisons of productivity levels. One is to find the appropriate conversion factor to express the value of each country's output in the currency of the other country. The second is to use consistent sources, and apply consistent concepts, in obtaining the basic figures for output and employment.

The conversion factor used in this study was based on a sample of unit values for ex-factory sales of comparable products between the two countries which were obtained from the Enquetes de Branches in France and the Quarterly Sales Inquity in the United Kingdom. Table 2 shows that in total only about 12 per cent of all manufacturing sales in France and 8 per cent of sales in the United Kingdom could be matched. This may seem low, and it certainly is in comparison with previous studies between Britain and Germany, the United States and the Netherlands (Smith, Hitchens and Davies 1982; Van Ark 1990). As in these previous studies matching was frequently hampered by different product specifications in the censuses of two countries. An additional problem for the present comparison was the decentralised way in which product information in France has been collected so that for many products only quantities and no unit values were published.(')

In order to reduce the risk that products which are overrepresented in the sample have an exceptionally high weight in the UVRs at branch and sector level, UVRs were aggregated in three steps: first, from the product level to the (four-digit) industry level (for example, footwear, paints or cement), then to the (two-digit) branch level (for example, wearing apparel, chemicals or stone, clay and glass products) and finally to the level of total manufacturing.(9) The aggregation from the product level to the industry level was made by using sales quantities as weights. The average unit value ratio for the industry is accepted if the individual UVRs are for products covering 30 per cent or more of sales in an industry. This requirement was met for 16 industries, which accounted for almost 20 per cent of manufacturing value added in each country. An industry which failed the 30 per cent test received the average UVR for all matched products in the branch to which it belongs. The UVRs for branches are derived by weighting the average industry UVRs by value added. When there were no UVRs at all in a branch, the average UVR of all other branches was applied. In the present comparison average branch UVRs were applied for food products and beverages, printing and publishing, and instruments and miscellaneous industries. Finally, the average UVR for manufacturing is obtained by weighting the average branch UVRs by value added.

On average, the UVRs for the ten branches for which matched products were available differed by some 13 per cent from the average UVR for total manufacturing.110) The estimates of UVRs for wearing apparel and electric engineering were particularly high, whereas those for paper products and stone, clay and glass products were relatively low. In addition the UVR for a large branch like food products and beverages, for which no products could be matched at all, was directly derived from the average UVR for the other branches. Even with the stage-wise aggregation procedure, there was therefore some reason to consider the validity of the assumption that unit value ratios for matched sales are representative for parts of non-matched sales in the sector. Table 3 presents estimates of the productivity ratios between France and the United Kingdom on the basis of alternative calculations of unit value ratios for branches. The second line of the table shows the productivity gap between French and British manufacturing if (1) the UVR for food products is assumed to be 13 per cent lower than estimated for this study; and (2) the relatively high UVRs for wearing apparel and electrical engineering are replaced by the average UVR of the other 'matched' branches. The productivity ratios in the third line of table 3 are based on the assumptions that (1) the UVR for food products and beverages is 13 per cent higher than the average branch UVR; and (2) the low UVRs for paper products and stone, glass and clay products are replaced by the average branch UVR. In fact, there is very little evidence to support the notion that the outlier' UVRs are wrong because of estimation errors.(") The 41 and 24 per cent estimates of the productivity gap between French and British manufacturing may therefore be taken as the upper and lower limits of the range in which the actual figure falls, and it is more likely to be near the 31 per cent estimate shown in the first line of table 1.

A crucial aspect of any comparison of product items between countries is to ensure that one is comparing product items of as similar a quality as possible. Today the 'potato-is-a-potato' rule is widely accepted, which implies that one can validly compare the price or unit value of two product items with broadly similar physical characteristics, irrespective of where, when and how it was produced. Quality problems are less frequently encountered in comparisons of ex-factory prices than in retail price surveys, because the former include more prices of intermediate products, such as steel, cement, planed wood or basic chemicals. These products, which are of a more homogeneous nature than consumer goods and investment goods, are by definition excluded from expenditure price surveys. However, comparisons of ex-factory prices derived from official sources, such as censuses of production, are sometimes criticised for their lack of specification in defining product qualities. In this study the original unit value ratios of six major products were thought to require an adjustment for quality differences between France and the United Kingdom. These products (cotton spinning and doubling, footwear, base paper products, paints, tractors and, most important, passenger cars) represented 56 per cent of all matched sales in France and 46 per cent of matched sales in the United Kingdom. The procedure involved the use of detailed information on the quantities and unit values of varieties of these major products from secondary sources. Quantities were mostly taken from trade sources, but it was difficult to derive ex-factory prices for different varieties. As an alternative, the ratio of retail prices or export unit values between different varieties within each country was taken to be representative for the (unknown) ratio of ex-factory prices. The ex-factory unit value for each variety was then calculated from these ratios and combined with the actual unit value for the product group as a whole, which was derived from the production census.

Although the effect of the quality adjustment was quite substantial in certain cases-for tractors and cotton spinning it was as much as 12 and 18 per cent of the original UVR respectively-the effect on the average UVR for the manufacturing sector as a whole was limited: the manufacturing UVR adjusted for quality differences was 10.47 francs to the pound compared to 10.20 francs for the unadjusted UVR. The most important effect on the UVR for manufacturing derived from the adjustment for passenger cars; the adjustment procedure for this product group is described in more detail in annex B.

It might be concluded that after the necessary adjustments, unit value ratios based on censuses of production (and related statistics) are superior to other conversion factors for sectoral productivity comparisons, because they relate most closely to the ex-factory prices of goods. Previous productivity comparisons between France and the United Kingdom have made use of the currency exchange rate, or at best an average ratio of retail prices (purchasing power parities). These are both somewhat higher than the UVR in this study: in 1984 the currency exchange rate was 11.68 francs to the pound; the PPP for a selection of manufactured goods was 11.94 francs to the pound in 1984. (12) One would therefore expect that previous comparisons of manufacturing productivity between France and the United Kingdom which are based on these higher alternative conversion factors would lead to a smaller productivity gap than the 31 per cent differential observed in this study. However, most previous studies have led to estimates of productivity gaps which are bigger than the ratio observed here.

Table 4 compares the Anglo-French productivity gap in manufacturing observed in previous studies with the estimate in this paper, after appropriate adjustments for differences in 'benchmark' years and after allowing for differences in annual person-hours in manufacturing.

The differences in the estimate of the productivity gap for 1984 based on these earlier comparisons and on the present study appear to be mainly due to their use of inconsistently defined output figures (taken from the national accounts) and labour input data (taken from the employment statistics). This contrasts with our approach based directly on production censuses which is preferable as the basic output and employment figures relate to precisely the same establishments. Gross value added per person employed in the United Kingdom was taken from the Report on the Census of Production. The French production census in fact consists of two separate sources, that is, the Enquete Annuelle d'Entreprise 1984 and the Enquetes de Branches 1984. Annex A describes how the gross value added per employee for manufacturing activities in France was derived by the amalgamation of these two sources.

3. The Anglo-French manufacturing productivity gap from 1973 to 1989: a review of the results

Table 1 in the previous section shows that the productivity ratios in some branches differed considerably from the average for total manufacturing in 1984. Two extreme examples were the paper and paper products branch and the wearing apparel industries. The substantial productivity gap in the paper industry was also observed in a detailed comparison plants, which showed that French plants produced 2.5 to 4 times as many tonnes of paper per man compared to British plants at the beginning of the 1980s. (13) The slow introduction of modern technology in the British paper industry was also highlighted in the Anglo-Dutch comparison of manufacturing productivity (Van Ark 1990, p. 76). The relatively high productivity in the British wearing apparel industry can be traced to the strong concentration on mass production of standardised items in this country'

It is worthwhile to check whether part of the productivity advantage in French manufacturing might be explained by a more favourable mix of manufacturing activities. If France were more specialised in activities with a high value added per worker, for example in manufacturing basic chemicals or electrical engineering products, the productivity level for the manufacturing sector as a whole could be higher than in the United Kingdom even if the productivity performance of individual branches was not so different between the two countries. In fact, the composition of the manufacturing sector in France and the United Kingdom is quite similar, although food processing has a slightly larger share in the British manufacturing sector and basic metals are more strongly represented in France (see table Al). Both countries are less involved in skill and technology-intensive industries which are characteristic of the manufacturing sectors in Japan, the USA and Germany (see, for example, OECD 1987).

Previous comparative studies have often looked at differences in productivity growth trends between countries. Between 1950 and 1973 output per person-hour in manufacturing in France increased at an average rate of about 6 per cent a year. If France's performance in this respect was not quite as impressive as in Germany (where the corresponding rate was close to 6.5 per cent a year), it was substantially above the average growth rate of 3.3 per cent in the United Kingdom during this period. These growth rates can be linked to the estimates of the productivity gap, as presented in the previous section, to derive the productivity gap between the two countries for other years than the benchmark' year. On this basis, it appears for example that around 1950 manufacturing productivity in France was about a third below the level in the United Kingdom, and that France had fully caught up with Britain by the mid-sixties.

Table 5 provides a more detailed picture of changes in the comparative productivity performance by branch in French and British manufacturing since 1973, the year when the first oil crisis-ended the 'golden age' of post-war growth. Although output per person-hour in French manufacturing as a whole was ahead of the United Kingdom by 13 per cent in 1973, some branches in British manufacturing then still showed higher productivity compared to France. By 1979, French manufacturers were performing significantly better than their British counterparts in almost all branches. The productivity gap was in particular large in basic metals and metal products and in mechanical engineering and transport equipment.

Since 1979 the productivity performance of the British manufacturing sector has improved relative to France. Output per person-hour increased at 4.7 per cent per year on average between 1979 and 1989, compared to only 3.2 per cent in France. As a consequence the manufacturing productivity gap between France and the United Kingdom narrowed from 142 per cent in 1979 to 122 per cent in 1989. It is not clear whether the relative productivity improvement in the United Kingdom reflects a fundamental change in performance, including better use of technology and improved labour relations, which can be maintained in the future. Firstly, the gap between France and the United Kingdom in 1989 remains greater than it was in 1973. Secondly, table 5 shows that the narrowing of the gap has not been of the same magnitude in all branches. This appears to be due mainly to the varying degree of productivity growth in British manufacturing branches. The average annual growth rates of gross value added per person-hour in basic metals and metal products and electrical engineering were 7 to 8 per cent between 1979 and 1988 compared with only 3 per cent in mechanical a engineering and transport equipment. As a consequence, Britain's relative productivity level in basic metals and metal products and electrical engineering greatly improved, whereas it remained almost unchanged for mechanical engineering and transport equipment. it can therefore be concluded that branchspecific' factors, such as the performance of particular firms or special market conditions in a particular branch are at least as important in explaining Britain's improved productivity performance as overall factors such as those discussed below.

The narrowing of the manufacturing productivity gap between the United Kingdom and other European countries such as the Netherlands and Germany since 1979 was shown in previous studies. In 1988 Dutch productivity was estimated at 44 per cent above the British level down from 69 per cent in 1979 (see Van Ark 1990). The gap between German and British manufacturing narrowed from 62 per cent in 1979 to 37 per cent in 1988.(") There are several possible reasons why the overall productivity performance in French manufacturing was not as good in comparison to the United Kingdom as that observed for the Netherlands and Germany. Both France and the United Kingdom have been slow in adjusting their industrial structure from traditional industries, such as textiles and steel, to innovative and capital intensive industries, for example engineering and chemicals. This is also reflected in the low skill-intensive nature of these country's exports (see Barker, Britton and Major 1984; Turpin 1989). Of course, one should not neglect the successful performance of some French firms in market niches such as nuclear energy and armaments, and of British firms in, for example, instrumental engineering, but the effect of these success stories on the overall performance of each country's manufacturing industry seems limited. By contrast, Germany in particular has continued to exploit the comparative advantage it had already in branches such as engineering and chemicals (OECD 1987). Another frequently cited weakness of French manufacturing industry concerns the relatively small average size of plants. This subject will be discussed in more detail in the next section.

4. Differences in plant-size between France and the United Kingdom

In 1988 the median employment size of a manufacturing plant in the United Kingdom (defined so that half of the labour force is employed in plants of a smaller size and the other half in plants of a bigger size) was some 60 per cent above the corresponding median plant size in France, that is, 240 against 150 employees.

It has been suggested that French manufacturing did not reap the benefits of economies of scale to their full extent. This argument has been used in supporting policies to develop and strengthen large firms in France. (17) However, it seems that to Germany, the United Kingdom did not benefit from a larger plant size in productivity terms, because relatively large plants were concentrated in branches with limited economies of scale. (18) This is confirmed in table 6, which shows that in 1988 the difference in median plant size between France and the United Kingdom was large in non-durable consumer good industries which includes, for example, wearing apparel, wood products and printing and publishing.

One expects greater scale advantages in capital-intensive industries, such as intermediate goods (in particular chemicals and basic metals) and machinery and equipment (in particular motor vehicle production). However, it seems that even in these industries the United Kingdom did not benefit greatly from large scale of operation, because of relatively unfavourable labour relations in large plants (Prais 1981a). Between 1979 and 1988 the share of the number of employees in the very large plants with more than 1,000 employees fell sharply from 36 per cent to 24 per cent of the total number of employees in these branches. Median plant size in France has now risen to 81 per cent of the plant size in the United Kingdom in these branches. The reduction of large plants seems to have contributed to Britain's improved productivity performance, which was particularly strong in these industries (see Section 3).

The reduction in median plant size in light industries between 1979 and 1988 was much smaller than in heavy industries in both France and the United Kingdom, and the relative plant size hardly converged between the two countries (table 6). The narrowing of the productivity gap was also less significant in these industries compared with heavy industries.

The largest difference in plant size occurred in the food processing industry, where half of the labour force in France is employed in plants with 70 or less employees and half of British workers are employed in plants with more than 350 employees. Although a great diversity exists, the British food processing industry is known for its steady increase in scale of operation, which in some industries dates back to as early as the beginning of this century. This process almost wiped out traditional artisan industries, such as craft bakeries and butchers in the United Kingdom. In France, artisans accounted for 15 to 20 per cent of total employment in food manufacturing plants in 1988. This high proportion could be related to the higher share of the agricultural sector in the French economy and the larger number of people still living in rural areas. On the basis of that assertion one could argue that small plants are still relics of preindustrial times, and that France's productivity record might benefit from an increase in average plant size. However, it must be noted that whereas within a single country output per employee in a larger plant may well be greater than that in a smaller plant, in comparison with another country the scale-factor can be easily outweighed by other factors, for example differences in the utilisation of the capital stock and the average level of skills of the labour force.(21)

5. Evaluation of the results

The estimates in this article show output per person-hour worked in the manufacturing sector of France to be 22 per cent above the corresponding level in the United Kingdom in 1989. This gap is substantially smaller than the 42 per cent difference estimated for 1979. However, the French productivity advantage over the United Kingdom remains greater than the 13 per cent observed at the end of the post-war golden age' of French industrial growth in 1973. Even if the faster growth rates of productivity in British manufacturing of the 1980s can be maintained, it will still take another 4 to 5 years before pre-1973 comparative level of productivity can be reached.

The manufacturing productivity gap between France and the United Kingdom is about half of that observed between Germany or the Netherlands and this country. One must be cautious in stressing one or two explanatory factors for these differences in productivity gaps. The present comparisons of the countries' production censuses are not all based on the same benchmark year and the same country weights. Moreover, the comparison in this article is based on a smaller sample of unit value ratios which were used to convert output to a common currency. Nevertheless a few, partly tentative, observations can be made on manufacturing productivity performance in the United Kingdom compared to three other important industrial nations in Europe.

Compared to Germany and the Netherlands, both France and the United Kingdom were handicapped by a more outdated industrial structure with little or no specialisation in innovative and capital intensive industries by the end of the 1970s. Both countries have also been relatively slow to adjust their industrial structure and expand trade in sectors where demand was strong (OECD 1987; Turpin 1989). However, Britain's relative improvement in productivity performance over France has been particularly strong in two major branches, that is, basic metals and metal products and electrical engineering.

This study looked also at differences and trends in average plant size in France and the United Kingdom. The substantially larger average plant size in the United Kingdom does not seem to have yielded much benefit in terms of productivity performance. Many of the larger plants in this country are located in light industries with generally limited economies of scale. In addition large plants in Britain have been more hampered by unfavourable labour relations, so that the sharp reduction in plant size in some heavy industries in the United Kingdom over the last decade has in fact helped to narrow the productivity gap between France and the United Kingdom. in the light industries, in particular food manufacturing, the median plant size in Britain is still comparatively large, whereas its relatively productivity performance is about average. However, it must be noted that there is no evidence of a very strong relationship between differences in average plant-sizes and productivity gaps between countries.

Unfavourable labour relations are sometimes mentioned as an important factor in explaining relatively low productivity levels in British manufacturing. This factor may also play a role in the comparison between France and the United Kingdom: French trade unions have never been as strong as their counterparts in the United Kingdom either in terms of membership numbers or in terms of their influence within firms (Hough 1982).

Previous cross-country comparisons by the National Institute between manufacturing plants of a similar size producing the same product emphasised the crucial importance of skill levels which contribute to the United Kingdom's relatively low level of labour productivity. Recent studies show, for example, that almost two thirds of the manufacturing labour force in the United Kingdom have no vocational qualification (although they may have completed some form of general education) compared to about half of employees in French manufacturing; in the Netherlands the unskilled section of the manufacturing labour force is not more than 40 per cent, and in Germany it is as low as 25 per cent. 122, These different numbers of vocationally qualified personnel in manufacturing show some correspondence to the productivity gaps observed in this article. However, detailed studies of matched plants' and comparisons of training and education are required to unravel the relative importance of skills among the other factors which account for the persisting productivity gaps between countries in western Europe (see, for example, Prais 1990).