Printer Friendly

Productivity levels in British and German manufacturing industry.

Relative levels of labour productivity are estimated to have been about 22 per cent higher in German than in British manufacturing in 1987. The German productivity advantage was most pronounced in non-electrical engineering, vehicles and metals. The British performance was relatively better in food, drink and tobacco and textiles and productivity levels appear to be about equal in the two countries in chemicals and electrical engineering. About 80 per cent of the productivity gap in aggregate manufacturing can be accounted for by differences in the levels of both physical and human capital. The aggregate productivity ratio of 22 per cent is lower than that found for 1968. The time Pattern of relative productivity in the intervening two decades shows a considerable increase in the 1970s followed by a rapid narrowing of the productivity gap in the 1980s.

1. Introduction

Following a decade of slow growth, labour productivity in British manufacturing industry increased considerably in the 1980s. There is little doubt that the British productivity performance in that decade was better than in most OECD countries (Matthews and Feinstein (1990)) but measuring the extent to which Britain has succeeded in narrowing the productivity gap with other industrialised countries requires information on the levels of labour productivity in British manufacturing relative to those in her major competitors. Previous issues of the Review (van Ark (1990a 1990b)) looked at productivity levels in British manufacturing relative to those in France and the Netherlands. This study considers the position relative to Western Germany.(1)

Output per worker-hour in German manufacturing is estimated in this article to have been about 22 per cent above that in the UK in 1987. This result is smaller than previous estimates, based on indirect secondary statistical sources, which put manufacturing labour productivity at over 40 per cent higher in Germany for the late-1980s.(2) Section two sets out the method used to arrive at the estimate for total manufacturing and examines the sensitivity of the result to variations in the underlying assumptions. Section three presents estimates of the productivity ratio for a number of industry branches within manufacturing.

Differences in the amount of fixed capital available to manufacturing in the two countries may account for some of the gap in labour productivity. Research at NIESR suggests differences in the skill levels of the work-force is an important factor determining relative productivity. Section four therefore attempts a measurement of the impact of differences in both physical and human capital on relative productivity in the two countries.

Time series data show that the productivity ratio widened considerably in the seventies, reaching a peak relative German advantage of about 50 per cent in 1980. The following decade witnessed a considerable narrowing of the gap so that by 1987 the relative productivity gap between the two countries was slightly lower than in the late-1960s. This was achieved at the expense of a huge reduction in UK manufacturing employment, most of which occurred after 1979. Section five considers the time series pattern of relative productivity.

2. The productivity ratio in aggregate manufacturing Levels of manufacturing labour productivity in Germany relative to those in Britain are measured as relative real gross value added divided by labour input. Published sources give data on nominal value added which need to be converted to a common currency. This implies that the measurement of relative productivity requires information on the relative prices of manufactured products.

The calculation of relative prices in this article employs information on quantities and sales of a range of manufactured products to estimate their unit values in each country. The ratio of German to UK unit values were calculated for as many manufactured products as was possible given data limitations (the fact that the sales value and/or quantity sold were not disclosed for all products) and the desire to ensure that the products were closely matched in the two countries. A weighted average of these unit value ratios' (UVRs) was then used as a conversion factor where the weights reflected the importance of products in matched sales at the industry level and the importance of each industry in total manufacturing. The price ratios for `matched products' are assumed to be representative of price ratios for non-matched products within each industry. The estimation of price ratios using unit values has a long history and was used, for example, in Paige and Bombach (1950), Smith, Hitchens and Davies (1982), Maddison and van Ark (1988) and van Ark (1990a, 1990b).

About 30 per cent of non-engineering manufactured sales were matched. In general, it was not possible to achieve matches in mechanical and instrument engineering and in most of electrical engineering so that, for these sectors, use was made of purchasing power parity prices, available in Eurostat (1988). Appendix table A.1 shows the percentage of matched sales by industry branch. Further details of the methodology used to construct relative prices are given in the appendix.

Of equal importance to the estimation of relative prices in measuring relative productivity is the necessity of ensuring that the output and employment data are comparable across the two countries. The censuses of production 13) were used as the primary data sources since both countries use the same concepts of gross value added and employment. Also, use of the censuses ensures that both the output and employment components of the productivity calculations come from returns from the same reporting units. Comparability required excluding small firms, employing less than twenty workers, from the UK census. Labour input is measured as numbers of workers times average annual hours worked. The appendix contains a discussion of the derivation of series on hours worked.

The results, for total manufacturing, together with some alternative estimates are shown in table 1. Comparing German and UK manufacturing, output per worker was estimated to have been about 13 per cent higher and output per worker-hour about 22 per cent higher in Germany in 1987. The corresponding productivity ratios for 1968 implied in Smith, Hitchens and Davies (1982) were 23 per cent and 26 per cent, respectively. Therefore, the gap in German to British output per worker appears to have narrowed significantly between 1968 and 1987 and the gap in output per hour worked has shown a small decline over the same period.

How sensitive are these estimates to the calculation of relative prices? An alternative conversion factor is to use the average manufacturing purchasing power parities (PPPs), published in Eurostat (1988), which are described in Summers and Heston (1988). These are based on price quotations for large numbers of products. PPPs are measures of relative retail prices and so will differ from producer prices in that they exclude the prices of semi-finished products and exports but include the prices of imported goods. The unit value ratios are closer in definition to producer prices and allow the estimation of productivity ratios in industries producing semi-finished products, hence our preferred use of UVRs. However, a large discrepancy between the aggregate manufacturing price ratio as measured by UVRs and PPPs could indicate that the former were not representative of relative German to UK prices in general. Table 1 shows that, in fact, the use of PPPs and UVRs imply similar productivity ratios.

A third alternative is to use the official exchange rate. This is generally not considered to be a reliable conversion factor as it tends to be overly influenced by short-term capital movements. The use of the exchange rate implied a productivity ratio which was over twenty percentage points above that using either PPPs or UVRs. The D-Mark/L exchange rate fell throughout 1986 and 1987 so the fact that both the PPP and UVR were significantly above the exchange rate in 1987 could be reflecting lags in the adjustment of relative prices to the exchange rate.

The UVR calculations do not, in general, adjust for differences in the average quality of products in the UK and Germany. The unit value ratios are calculated for fairly detailed product specifications such as number of women's blouses but, in general, cannot distinguish quality variations in the products. A high average quality of manufactured products in Germany relative to those in the UK would impart an upward bias in relative prices and hence imply a higher productivity gap than that given in table 1.

There is some evidence that German goods are of a higher average quality in such diverse goods as kitchen furniture (Steedman and Wagner (1987)), women's clothing (Steedman and Wagner (1987)) and biscuits (work in progress at NIESR). It is difficult to know how representative these product areas are of manufactured goods in general. However, there are some reasons for supposing that the difference in quality may not be so important at the level of total manufacturing.

The first is the fact that the PPPs and UVRs produce similar results since the measurement of the former, based on very large samples of matched products, is designed to account for differences in quality. However, in the PPP calculations, investigators in different countries are working to a common product specification but there is no direct comparison of products so it is possible that the PPPs also do not adequately account for quality differences.

The second reason for doubting the importance of quality differences is the time consistency of our relative price calculation with that given in Smith, Hitchens and Davies (1982). Using the producer price indexes in both countries to update the Smith et al 1968 UVR gives a price ratio of 3-59 D-Mark to the L in 1987 which is remarkably close to the 1987 benchmark estimate of 3-5 D-Mark to the L shown in table 1. Both benchmark estimates are based on (slightly different) samples of products but the producer price indexes cover all manufacturing and adjust for changes in quality over time. If German production had moved to higher quality products in the past two decades then we would not expect such a correspondence between the 1987 benchmark and that implied by the time series extrapolation. This argument does not, however, preclude a constant quality differential.

The fact that the UVR in 1987 is some 20 per cent above the exchange rate should not be taken as evidence of superior German quality since projecting the UVR forward to 1989 and 1991, using producer price indexes, gives price ratios which are about 9 per cent and 5 per cent, respectively, above the exchange rates for those years. Hence the amount by which the UVR exceeds the exchange rate has declined over time suggesting that the `high' UVR in 1987 was indeed a short-run phenomenon.

The above discussion suggests that the quality bias may not be significant. The arguments are based, however, on indirect evidence. Only direct evidence on quality differences can show if German products are much superior to British products. Much more work needs to be done to resolve this issue. Since no information exists on the extent of quality differences for most manufactured goods it is not possible, in general, to adjust for this. The exception was an adjustment in passenger cars and trucks which took account of differences in cylinder capacity using the method outlined in Van Ark (1990b); this made little difference to the UVR for motor vehicles.

Although much of the originality of the estimates in this paper depend on the price calculations, it turns out that differences between this and alternative estimates hinge more on the measurement of output and labour input. The most soundly based alternative estimate is that of Hooper and Larin (1988) who presented a figure of a 46 per cent relative German productivity advantage in 1987. Hooper and Larin used manufacturing PPPs as conversion factors which were shown above to yield similar estimates to those based on UVRs.

Hooper and Larin's primary data source are series supplied to them by the US Bureau of Labour Statistics (BLS) which takes output and employment from individual country's national accounts. In Germany, the national accounts figures for gross value added and employment come primarily from the production census and so their use makes little difference to the productivity ratio. In the UK, gross value added in the national accounts is based on measures of income and comes from tax returns from the Inland Revenue whereas employment comes from the 'Census of Employment' produced by the Department of Employment.

National accounts value added was 6 per cent lower and national accounts employment was 7 per cent higher than the corresponding census estimates in the UK in 1987. The magnitude of these discrepancies between the national accounts and the censuses of production tends to be fairly constant over time. The use of UK national accounts sources would add about 15 percentage points to the German/UK productivity gap in 1987. We believe an adjustment to a national accounts basis is not valid since the fact that output and employment come from very different sources and are produced by different statistical offices means that it is doubtful if they are compatible in the sense of referring to the same establishments. in the national accounts it is difficult to disentangle manufacturing and distribution since many companies are engaged in both sectors (see CSO (1985) p. 35). The advantage of using the census of production is that the figures for value added and employment refer to the same reporting unit. There is no reason to suppose that the census of production systematically under records employment or overstates output.

Hooper and Larin use the BLS estimates of average annual hours worked, which adds about 7 percentage points to the productivity ratio. The main difference between the BLS average hours series and that used in the current paper occurs for the UK where the former is about 8 per cent higher; there is little difference between the two estimates for Germany. In constructing UK average hours worked, the current study employed information on weekly hours worked which distinguished manual and non-manual workers, males and females, full-time and part-time, and on numbers of weeks holidays and time lost due to sickness, maternity leave and strikes. This calculation was carried out for individual industries using both published and unpublished sources primarily from the Department of Employment and the Labour Force Survey. The result was an estimate of average weekly hours equal to 40-4 and weeks worked per year of 43-6. We cannot compare our methodology with that used by the BLS 141 but we are reasonably confident in the accuracy of our result.

Further evidence in support of a 20 per cent rather than a 40 per cent productivity gap is that the former appears to be more reasonable when compared with estimates for the aggregate economy which show little difference between Germany and Britain. Summers and Heston (1991) estimate that 1988 GDP per worker in Germany was only 6 per cent higher than in the UK and Maddison (1989) calculates GDP per worker hour to have been 5 per cent higher in Germany in 1986. if output per worker-hour in German manufacturing had been 40 per cent higher than in the UK in 1987, relative productivity in all other sectors would have been about 7 per cent higher in the UK. This implication for productivity in non-manufacturing led Feinstein (1988) to question the plausibility of some previous estimates of the German/UK productivity ratio. The manufacturing productivity gap of 22 per cent between Germany and the UK therefore seem reasonable on the basis of the whole economy productivity calculations.

Finally, in this section, a comparison of the benchmarks for 1968 and 1987, shown in table 1, implies that the ratio of German to UK output per worker was significantly lower in 1987 than in 1968. The reduction in the productivity gap was smaller once adjustments were made for differences in the trend in average hours, which reflected primarily the relatively large increase in number of weeks paid holidays in Germany. On either measure, UK manufacturing appears to have more than compensated for its poor performance in the period 1973 to 1979. Backdating the 1987 benchmark using time series data for both countries yields a productivity ratio which is within two percentage points of the 1968 benchmark estimate of Smith, Hitchens and Davies (1982).[5] Hence the current estimate is consistent over time with previous results, which further strengthens its plausibility. An examination of the time pattern of relative productivity is provided in section five below.

3. The sectoral pattern of relative productivity The productivity ratio for aggregate manufacturing was built up from estimates for over fifty individual industries. This section summarises the estimates for fourteen sectors within manufacturing; results are shown in table

2. Space limitations preclude the presentation of the more detailed results which are set out in O'Mahony (1991). Appendix tables A-2 and A-3 show the levels of productivity by sector for each country both on a per worker and a per-worker hour basis. This section includes a brief analysis of some reasons for the cross sector variations but the small number of observations mean that the results should only be seen as suggestive.

Looking first at output per worker, the UK appeared to have a productivity advantage over Germany in 1987 in the manufacture of basic metals, chemicals and electrical engineering. The remaining eleven sectors show a German productivity advantage. The timber and paper sectors show the highest ratio of German to UK productivity but the above average productivity ratio in non-electrical engineering and vehicles are of greater importance since these sectors represent a large share of output in both countries. The very high German productivity advantage in paper and board may seem surprising but is consistent with the findings of van Ark (1990a) for the Netherlands.

The picture is slightly different when productivity is measured by output per worker-hour. The greater number of hours worked in British manufacturing is most pronounced in engineering, vehicles and metals implying an even greater productivity gap than on a per worker basis. This is due to the above average incidence of overtime in these sectors in the UK. The adjustment for differences in hours worked leads to a slight German productivity advantage in electrical engineering and basic metals but it lowers the productivity ratio in food, drink and tobacco and textiles. Relative productivity levels in the chemicals industry, on a per hour basis, appear to be very similar in the two countries.

The final column of table 2 above shows the 1968 ratios of output per worker from Smith, Hitchens and Davies (1982)."' In most sectors the productivity ratio declined over the period 1968 to 1987 but there were some exceptions, notably food, drink and tobacco, clothing and paper. The greatest improvements in relative British performance occurred in basic metals, non-metallic mineral products, chemicals, vehicles and textiles. All showed substantial reductions in employment in the UK in the 1980s. The correlation between the change in relative productivity between 1968 and 1987 and change in UK employment between 1979 and 1982 (taken from the production censuses for those years) was negative and significant at -0-63 so that sectors with the greatest reduction in UK employment in the eighties appeared to experience the highest relative productivity improvement. This is consistent with the shock effects of the 1980 recession discussed by Layard and Nickell (1989).

Is there evidence of convergence of productivity levels across industries? The correlation between the change in the German to UK productivity ratio from 1968 to 1987 and the level of German to UK productivity in 1968 was found to be negative and significant at 0-58. Therefore, there is some evidence that the sectors which showed the greatest UK relative productivity improvement were those whose productivity levels in 1968 were furthest from those in Germany.

Table 3 shows the shares of manufacturing employment and sales accounted for by each of the sectors in 1987 and the corresponding employment shares in 1968. in 1987 German manufacturing employment was more concentrated in heavy industries such as vehicles and engineering, and chemicals, whereas UK manufacturing employment was more concentrated in light sectors such as food, textiles and clothing. This was a change from the industrial structure in 1968, which was similar in the two countries. In both countries, the share of sales accounted for by chemicals and food is greater than their shares of employment.

Do these differences in the structure of manufacturing employment in the two countries help explain the level of the productivity gap in 1987? We first calculated the ratio of output per worker if the shares of employment in both countries were equal; this yielded productivity ratios of 118.8 and 118-5, respectively, using UK and German employment weights. Thus about three percentage points of the twenty-two percentage point productivity gap, in 1987 aggregate manufacturing, was due to higher German employment shares in sectors with above average levels of labour productivity.

Are structural factors important in explaining the change in the productivity gap between 1968 and 1987?(7) This will depend not only on relative sectoral growth rates but also movements between capital intensive high productivity level sectors such as chemicals and those with lower levels such as textiles. The change in relative productivity was decomposed into a component due to changes in relative productivity growth rates at fixed weights and one due to changing weights taking account of differences in productivity levels across sectors. The decomposition is taken from Nordhaus (1972); details are given in O'Mahony (1991).

The result was that in the absence of structural change the ratio of German to UK output per head would have been about three percentage points lower in 1987. Employment shares in sectors with high growth rates, on average, tended to increase in Germany; this was most pronounced in electrical engineering and vehicles. In the UK there appeared to be no association between high growth rates and increases in employment shares. In both countries, on average, employment shares increased in sectors with above average productivity levels but the effect was stronger in Germany. In both, the high productivity level chemicals sector increased its share and the low productivity level clothing sector in Germany and textile industry in the UK experienced declining weights.

Changes in industry structure in the two countries, therefore, led to a smaller decline in the productivity gap than would have been the case if these structural factors had not been present. Greater UK productivity growth rates in the majority of sectors, however, remains the primary source of the improved relative British performance.

We next examine the distribution of average plant size across sectors in the two countries; this information is summarised in table 4. The measure of average plant size is the median of the distribution of plant employment, that is, it is the plant size for which 50 per cent of the workforce are employed in plants with more than that number of employees. In 1987, the median plant size for all manufacturing was some 30 per cent lower in the UK than in Germany. Prais (1981a) showed that the median size in Germany was about 7 per cent lower than in the UK in the period 1970-3. In both countries the median appeared to fall since the Prais study but the decline was greater in Britain, about 40 per cent as against a fall of less than 20 per cent in Germany.

The median German plant is considerably larger in the heavy sectors such as engineering, metals and vehicles whereas the median UK plant is larger in lighter sectors such as textiles, clothing and, most strikingly, in food, drink and tobacco. This was also found to be the case by Prais for 1970-3 but his calculations were carried out for some 30 industries whereas the current estimates are for just fourteen sectors so comparisons between the results for the two periods are difficult. The correlation coefficient between relative productivity and relative median plant size was found to be insignificant at 0-04 for the sample of fourteen sectors.

Are the sectors where relative UK productivity improved the least over the period 1968-87 those with relatively larger German plants? The correlation between the productivity improvement in the UK relative to Germany and the ratio of German to UK plant sizes in 1987 was found to be significantly positive at .65. Therefore, the greatest relative improvement occurred in sectors where the UK had a size disadvantage compared to Germany in 1987.

This result may seem surprising at first sight but it could merely be reflecting the changes in UK manufacturing which followed the 1980 recession. The median size of UK plants has shown a considerable decline in the 1980s(8) reflecting both a scaling down of existing operations and increased competition. The sectors with the highest ratio of German to UK plant size could be those in which the increased competition in the UK was most pronounced or where the reduction in employment in the 1980s led to a large reduction in average plant size.

4. The contribution of physical and human capital International comparisons of physical capital stocks present even more difficult measurement problems than the calculation of relative levels of output and labour input. The statistical offices, in both countries, base their capital stock estimates on the perpetual inventory method, which cumulates past investments and subtracts retirements, but they employ different assumptions on the average service lives of assets and the form of the retirement distribution. It is difficult. therefore, to disentangle how much of the relative levels of capital, as shown in official publications, is due to true investment in physical capital and how much is due to methodological differences.

A common approach to resolving this comparability problem is to assume the same service lives in the two countries-this is the method adopted by Maddison (1991) and Summers and Heston (1991). The estimates of relative levels of capital, presented in table 5 below, assume service lives of about 40 years for buildings and 15 years for equipment and vehicles and are about equal to the service lives used by the Statistisches Bundesamt. These were chosen, rather than the longer British lives of 60 years for buildings and over 20 years for machinery and vehicles as the German assumptions are closer to those used by statistical offices in France and the United States. Gross capital stocks were calculated using the perpetual inventory method and assets were assumed to be retired uniformly between 20 per cent below and 20 per cent above the average service life. The German capital stocks were converted to L, using the 1985 purchasing power parities for machinery and industrial buildings given in Eurostat (1988).(9)

Table 5 shows that the ratio of capital to labour input was 38 per cent higher in German than in British manufacturing in 1987 so a significant proportion of the productivity gap should be due to this greater German capital intensity. If the production function is assumed to be Cobb-Douglas and its parameters are assumed to be the same in the two countries, then relative total factor productivity in the two countries, RTFP, can be estimated using the following equation:

RTFP = [ln(Y.sub.G./Y.sub.U)] - [Alpha.l.n(L.sub.G/L.sub.U) +

(1-[alpha]) ln (K.sub.G/K.sub.U)] (1)

where Y, 1, and K denote real output, labour and capital, respectively, oz is the average of the two countries' shares of wage payments in value added and G and U refer to Germany and the UK, respectively. This can be rewritten as:

[ln(Y.sub.G/Y.sub.U)-ln(L.sub.G/L.sub.U)] =

RTFP+(1-[alpha]) ln~(K.sub.G/L.sub.G) /

(K.sub.U /L.sub.U)] (2)

so that RTFP measures the residual relative labour productivity when account is taken of differences in capital intensity in the two countries The average value of [alpha] in 1987 was 0.75. The final line of table 5 shows that, in 1987, total factor productivity was 13 per cent higher in German than in UK manufacturing.

This calculation should only be taken as a rough approximation of the impact of differences in capital intensity since the calculation of relative capital stock levels is fairly crude. It could be argued that the service lives of British assets are indeed longer than in other industrial countries, in particular for industrial buildings which are not internationally transferable. If the service lives for buildings are allowed to differ in the two countries then the UK total gross capital stock rises by 8 per cent and the relative total factor productivity increases by two percentage points. Against this, the estimates in table 5 do not make any allowance for premature scrapping of assets in Britain after the two oil shocks which, as suggested by Muellbauer (1991), may have been substantial.

Finally the assumptions underlying the specification in equation (1) are restrictive. These are that the production function exhibits constant returns to scale, there is perfect competition in product and input markets and that the parameters of the production function are the same both across firms and countries.

Evidence for the existence of differential levels of skill in the manufacturing labour force in the two countries is contained in a series of studies carried out at NIESR (for example, Prais (1981b), Daly, Hitchens and Wagner (1985) Steedman and Wagner (1987), (1989) and Prais and Wagner (1988)). Table 6 shows the proportions of the workforce with qualifications at four levels, higher level, upper intermediate, lower intermediate and unskilled for 1987. It shows that there is little difference between the proportions with skills at the degree level but that British manufacturing has considerably lower proportions of workers with intermediate level skills; this confirms the differences found by Prais (1981) for the late-1970s.

The NIESR studies suggest that these difference in intermediate skills have a significant impact on productivity differences in the two countries. The skills embodied in the German workforce, based on formal training and external examinations, are more transferable between jobs within the same industry and between industries thus leading to a more flexible workplace. This flexibility is most important in periods of rapid technical change. The difference in skill levels enables German manufacturers to minimise the effects of machinery breakdowns thus ensuring delivery on time and allows the more rapid introduction of new technology.

A standard method of incorporating different qualities of labour in the relative productivity calculation is to measure labour in effective, rather than natural, units. Suppose there are two types of labour, skilled, s, and nonskilled, n, where one unit of skilled labour is [lambda], times more productive than unskilled labour. Let L denote the labour force measured in natural units, for example number of employees. Then effective labour, L , can be written as:

L* = [Lambda L.sup S] + [L.sup.N]

A common method used to estimate the relative effectiveness of different qualities is to use information on their relative wages, that is, set

[Lambda] = [W.sup.S]/[W.sup.N]

where W denotes wage rates. Substituting effective units for natural units in equation (2) gives

ln(Y.sub.G/Y.sub.U) - In (L.sub.G/L.sub.U) =

RTFP + (1 - alpha) ln [(K.sub.G/L.sub.G)/ (K.sub.U/L.sub.U)] + alpha.ln (H.sub.G/H.sub.U) (3)


[H.sub.j] = [W.sup.s.sub.j/W.sup.n.sub.j)(L.sup.s.sub.j) + (L.sup.n.sub.j/L.sub.j)] = j, G, U

Equation (3) was implemented using the four types of labour and the relative skilled to unskilled wage rates shown in table 6 above; the latter were derived using the wage categories used in Prais and Wagner (1988), updated to 1987. The results were not found to be overly sensitive to variations in the wage categories used. It is important to emphasise that this calculation does not employ the absolute levels of wage differences in the two countries, which it could be argued are the result of, rather than cause, productivity differences, but rather each country's ratios of skilled to unskilled wage rates.

The ratio of effective to natural labour was found to be 1.11 in the UK and 1.23 in Germany. Labour productivity in effective units was calculated to be 109-8 implying that 12 percentage points or 55 per cent of the labour productivity gap is due to the differences in labour force skills. When combined with physical capital, using equation (3) above, relative total factor productivity is calculated to be 104-0. Therefore, differences in both human and physical capital explain over 80 per cent of the productivity gap in 1987.

It should be emphasised that this calculation is crude and is based on somewhat unreliable data. It is very difficult to estimate differences in levels of physical capital and to compare qualification levels across countries. Relative wage rates are based on occupation rather than skill groups and the wage rates may not fully reflect the effects of skill differences on marginal products; this is particularly the case for the UK where wages are likely to be affected by the degree of unionisation in different occupations. Nevertheless, the estimates do point to a crucial role for both physical and human capital in explaining differences between German and UK manufacturing productivity.

This growth accounting' method suggests that differences in physical and human capital have about equal weight in explaining the ratio of German to British labour productivity. The calculation, ignores the possible interaction between these two types of capital; it is likely that more physical capital requires greater amounts of workforce skills in its operation. Human capital as measured above is embodied in the labour force. Theoretical models of economic growth such as Lucas (1988), whose engine of growth is unlimited accumulation of human capital, emphasise the effects of differences in the 'stock of knowledge' which is not specific to any individual. in these models an additional term measuring externalities from human capital accumulation is included in the production function so that differences in skills have a greater relative impact than allowed for in the growth accounting method.

5. The time pattern of relative productivity Up to this point the discussion has focused on the relative levels of output per unit of labour input in the benchmark years, 1968 and 1987. This section examines time series for the productivity gap in the intervening period and in the years since 1987 and discusses some possible explanations of the observed pattern. Space limitations confine attention to time series for aggregate manufacturing.

The reduction in the ratio of German to UK productivity between 1968 and 1987, shown in table 1 above, did not occur continuously but rather the productivity gap increased throughout the seventies, reaching a peak of about 50 per cent in 1979, and then narrowed rapidly in the 1980s. This is illustrated in chart 1 which uses time series on real output and employment from the national accounts in both countries; the data series are shown in Appendix table A-4.

The reduction in the productivity ratio in the 1980s was due primarily to the sustained high growth rates of UK labour productivity in that period but was also partly due to a slowing down of German productivity growth rates. The reduction in the productivity gap continued up to 1989, but it increased again in 1990, when the ratios of output per person and per person-hour were estimated at about 10 per cent and 19 per cent, respectively. These projections to 1990, based on national accounts sources, may be subject to later revisions.

Table 7 shows the growth rates of productivity, on both a per worker and per worker-hour basis, in the two countries for selected time periods. The pronounced relative improvement in the UK productivity performance in the 1980s needs to be evaluated in the light of what happened to both output and employment. The period 1968-87 witnessed both an increase in real manufacturing output and a decline in employment- however, the magnitudes of these changes were very different in the two countries.

German manufacturing increased its output by about 40 per cent but decreased its employment by 600,000 workers or 7 per cent between 1968 and 1987. On the other hand, UK manufacturing output increased by a little over 10 per cent but the reduction in employment was 3 million workers or 37 per cent over the same period. Measured by numbers employed, the manufacturing sector was roughly the same size in the two countries in 1968 but it was over 25 per cent lower in the UK by 1987. Thus output growth in the UK was virtually stagnant for most of the period so that the improved productivity was primarily a result of shedding labour, most of which occurred in the 1980s employment declined by about 900,000 in the period 1968-79 and by about 2-1 million in the decade between 1979 and 1989).

German manufacturing experienced almost uninterrupted growth in output and a considerably smaller reduction in employment. The growth rates of manufacturing labour productivity in Germany in the 1980s, however, appear to have slowed relative to the growth rates experienced in both the 1960s and 1970s. Darby and Wren-Lewis (1990) estimated stochastic trends for labour productivity growth rates in a number of OECD countries. The trend for Germany was persistently downward, in contrast to most other countries whose trend was either constant or showed a slight recovery in the 1980s. Therefore, some of the explanation for the narrowing of the Germany/UK productivity gap in the 1980s lies in the reasons behind the German slowdown.

General literature on the German economy in the 1980s, such as OECD (1985, 1989) suggest that the primary source of the slowdown in German productivity has been historically low proportions of output which went to investment. A gap between real labour costs and productivity opened up following the first oil price shock and persisted to the mid-1980s (OECD, 1985). This in turn led to lower profits and hence reduced investment. The slowdown in investment led to an aging of the capital stock and hence lower growth rates of labour productivity.

Chart 2 shows the time pattern of the capital-labour ratios in both countries, where the 1987 relative levels are taken from table 5 above. There appears to have been a widening of the relative (German to UK) capital-labour ratios in the 1970s and a considerable narrowing in the 1980s, most of which occurred in the early part of that decade. The growth rate of the manufacturing capital-labour ratio in Germany declined from 4-2 per cent per annum in the period 1968 to 1979 to 2.3 per cent per annum in the years 1979 to 1988. The growth rates of the capita labour ratio in British manufacturing also declined in the latter period but the decrease was not so pronounced, from 4-7 per cent to 3.8 per cent per annum.

When allowance is made for differences in capital intensity in the two countries, the basic pattern of a widening of the productivity ratio in the 1970s followed by a substantial narrowing in the 1980s remains. This is shown in table 8 which presents total factor productivity growth rates in each country for selected time periods where labour's share is assumed to be constant at .75. These differences in the relative capital stocks take no account of premature scrapping of capital in the UK which, as has been argued by Muellbauer (1991), could have been substantial in the early 1980s. If this is important, then the growth of UK capital would have been lower arid the relative increase in total factor productivity greater in the period 1979-88 than shown in table 8.

What determined the very different patterns in relative productivity in the 1970s and 1980s? Matthews and Feinstein (1990) emphasise the importance of catch-up created by the gap in productivity levels between the UK and her main competitors, in explaining the performance of the British manufacturing sector in the 1980s. This argument appeals to the convergence models of Abramovitz (1986) and Baumol (1986) which suggest that countries producing far inside the technological frontier have more scope to implement innovations already in use in the productivity leaders so that productivity growth tends to be faster in countries with relatively low initial productivity levels. Some empirical evidence for GDP per worker growth rates in OECD countries is provided in Dowrick and Nguyen (1989).

German productivity levels were substantially above those in the UK from the end of the 1960s and throughout the 1970s. Thus there were large unexploited gains to UK manufacturing at the beginning of the 1980s which were rapidly implemented in the following decade. The catch-up model suggests that German manufacturing had much less scope for introducing new technology and hence its productivity growth in the 1980s was considerably slower. However, manufacturing productivity levels in the United States, the world productivity leader, were still some 40 per cent higher than those in Germany in 1980 and this gap widened in the subsequent decade (van Ark (1990c)). Hence the German slowdown is unlikely to be primarily due to that country operating close to the technology frontier.

Restrictive practices in both the product and labour markets help explain why British manufacturing failed to even begin the process of catch-up to German productivity levels in the 1970s. On the output side, the historical lack of competitive pressures has been well documented (for example, Broadberry and Crafts (1990, Crafts (1988). UK membership of the EC should have provided the competitive pressure for change but goverment's willingness to bale out firms in the 1970s provided a cushion against this process. Metcalf (1989) states that 'handouts to companies with product market difficulties reached an all time record' in the 1970s. Significant change on the product market side appeared To occur only after the major disruption of the 1980 recession.

The detrimental effects of restrictive practices, such as overmanning, in the UK labour market particularly in large plants with multiple unions, is well documented, for example, Davies and Caves (1987), Prais (1981). The problems appear to have been more severe in the 1970s despite attempts to reform the industrial relations system. Metcalf (1989) concludes that the economic environment in the 1970s was not conducive to the attempts to alter workplace procedures. This was due not only to uncompetitive product markets but also to the fact that incomes policies put negotiations over productivity levels off the bargaining agenda.

The 1980s productivity revival in UK manufacturing has been the subject of much research. There appears little doubt that the 1980 recession had a major shock effect in both the product and labour markets (Layard and Nickell (1989), Metcalf (1989)). A reduction in restrictive practices in the labour market would appear to have had a major input into the UK productivity revival in the 1980s (Bean and Symons (1989), Oulton (1990)). Increased competitive pressure in product markets appears to have been crucial in facilitating this process of change Machin and Wadhwani (1989), Brown and Wadhwani (1990), Haskel (1991)). Thus the 1980s witnessed an altering of the role of interest groups in the UK. The scene was then set for the process of catch-up to Germany productivity levels to take off.

The discussion above suggests that that the slowdown in German productivity growth in the 1980s may be related to historically low rates of capital formation but that total factor productivity growth also showed a considerable decline over its values in the late 1960s and the 1970s. Some German economists, for example, contributors to Fels and von Furstenburg (1989) suggest that, at a time when the UK economy was freeing itself from the worst forms of restrictive practices, the German economy was becoming ever more restrictive. Emerson (1988) presents evidence that labour markets are somewhat more restrictive in Germany relative to the UK. The cost of redundancy payments are higher in Germany and there are more legal obstacles to firing workers than is the case in Britain. Dismissal tribunals generally reach decisions which are favourably disposed towards the workers,(10) in contrast to UK tribunals where only about one third of cases are successful. The employment of temporary and fixed-contract workers also appears more difficult in Germany although there has been some liberalisation on this since 1985.

The impact of greater ob security is not necessarily negative, for example, Emerson (1988) argues that increased job security may provide incentives for firms to invest in more training of their workforce, and thus facilitate the introduction of new technology. However, it is possible that labour market rigidities may have had their most negative impact on productivity growth in the early 1980s. At that time German manufacturing output declined for the first time in two decades and firms may have been unable to shed as much labour as they desired. German manufacturing experienced zero growth in labour productivity between 1979 and 1982. The high cost of firing may have induced German manufacturing to delay the introduction of new technology during recessionary periods, since it could not recoup the costs in lower manning levels, and hence this may have contributed to the investment slowdown.

The change in German labour market legislation in 1985, which allowed firms to employ more workers on short-term contracts, should remove some of the labour market rigidities. The introduction of the single market should remove any distortions in the product market, for example, the elimination of subsidies to declining industries and will allow more mergers and take-overs by foreign firms. Therefore, any adverse effects of too much regulation are likely to be lessened in the future. In this respect, it is interesting that German productivity growth rates from 1988 appeared to return to levels experienced in the 1970s.

5. Conclusion

This paper has estimated that, in the late-1980s, German manufacturing productivity levels were over 20 per cent higher than in Britain This productivity gap, which is more realistic than previous estimates when compared with relative GDP productivity levels, remains substantial. The historical pattern of the productivity ratio showed a widening of the gap throughout the 1960s and 1970s, with a particularly poor relative UK performance in the period 1973-79, followed by a rapid narrowing in the 1980s. The level of the productivity gap is now slightly below what it was in the late-1960s. This was achieved at a cost of a reduction of about a third in the UK manufacturing labour force.

The ratio of German to British labour productivity increased in 1990 and preliminary productivity estimates for 1991 show German growth rates in manufacturing to have been above those in the UK for that year. Therefore, in the short run the productivity gap appears to have widened again. What are the prospects in the long run? The results in section four of this paper suggested that differences across the two countries in stocks of physical and human capital could explain much of the 1987 gap in levels of output per worker-hour between Germany and the United Kingdom. Reducing this disparity in levels of capital would therefore appear to be important if the productivity gap is to be narrowed further.

Crafts (1991) suggests that the improved relative UK position was due primarily to more efficient use of factor inputs in that decade and not to any greater investment in physical capital, human capital or research and development. Section five of this paper confirmed that the growth in UK manufacturing capital per unit of labour input was in fact lower in the 1980s than in the previous decade. There has been little change in the relative stocks of skilled workers since the mid-1970s. Prais (1981) shows the proportions of the workforce in manufacturing with qualifications at higher and intermediate levels in the mid to late-1970s. Both countries have shown a reduction in the proportion of the workforce with no qualifications. There has been a doubling of the proportions of the labour force with higher level qualifications in each country but the difference in intermediate qualifications between Germany and Britain has not changed significantly. Recent theoretical contributions to explaining divergences across countries in economic growth place particular emphasis on differences in levels of human capital. Redressing the skill deficiency would therefore appear to be important if British manufacturing is to get close to the productivity levels now enjoyed in Germany.


Abramoviz, Moses. (1985) Catching up, forging ahead, and falling behind',Journal of Economic History, vol. 46, pp. 385-406.

Baumol, William J. (1986) Productivity growth, convergence and welfare: what the long-run data show', American Economic

Review, vol. 76, no. 5, pp.. 1072-85.

Bean, C. and Symons, J. (1990) Ten years of Mrs T.', Centre for Labour Economics, Discussion paper no. 370, London School of


Broadberry, S.N. and Crafts, N.F. R. (1 990) 'Explaining Anglo-American productivity differences in the mid-twentieth century',

Oxford Bulletin of Economics and Statistics, vol. 52, no. 4, pp. 375-402.

Brown, W. and Wadhwani, S. (1990) The economic effects of industrial relations legislation since 1979' National Institute

Economic Review, February.

Central Statistical Office (1985) United Kingdom National Accounts: Sources and Methods', third edition, HMSO, London.

Crafts, N. (1988) British economic growth before and after 1979: A Review of the evidence', CEPR discussion paper no. 292.

Crafts, N. (1991) Reversing relative economic decline? the 1980s in historical perspective', Oxford Review of Economic Policy,

vol. 7, no. 3, p. 81-98.

Daly, A. Hitchens, D.M.W.N. and Wagner, K, (1985) Productivity, machinery and skills in a sample of British and German

manufacturing plants', National Institute Economic Review, February.

Darby, J and Wren-Lewis, S. (1991) Changing trends in international manufacturing productivity', University of Strathclyde,

forthcoming, Scandinavian journal of Economics.

Davies, S. and Caves, R. (1987) Britain's Productivity Gap, Cambridge University Press.

Dowrick, Steve and Nguyen, Duc-Tho. (1989) 'OECD comparative economic growth 1950-85: catch-up and convergence',

American Economic Review, vol.79, no.5, pp.. 101 1-30.

Emerson, M. (1988) Regulation or deregulation of the labour market: Policy regimes for the recruitment and dismissal of

employees in the industrialised countries', European Economic Review, vol. 32, no.4, pp.775-817.

Eurostat (1988) Purchasing power parities and gross domestic product in real term: results 1985', Statistical Office of the

European Communities, Luxembourg.

Feinstein, C. (1988), 'Economic growth since 1870: Britain's performance in international perspective', Oxford Review of

Economic Policy, vol.4, no. 1, pp. 1-13.

Fels, Gerhard and von Furstenburg (eds) (1989), A Supply-Side Agenda for Germany, Springer-Verlag, Germany.

Hellwig, M. and Neumann, M. (1987),'Economic Policy in Germany: was there a turnabout?', Economic Policy, vol. 5, October.

Hitchens, D.M.W.N, Wagner, K, and Birnie, J.E. (1990), Closing the Productivity Gap: a comparison of Northern Ireland, the

Republic of Ireland, Britain and West Germany, Avebury, Gower publishing Company Limited, England.

Hooper, P. and Larin. K.A. (1989), 'International comparisons of labour costs in manufacturing', The Review of Income and

Wealth, series 35, no. 4, December, pp. 335-356.

Layard, R. and Nickell, S. 1989),'The Thatcher miracle', American Economic Review, May, pp. 215-219.

Lucas, R.E. (1988), 'The mechanics of economic development', Journal of Monetary Economics, vol.22, pp. 3-42.

Maddison, A. and van Ark, B. (1988), Comparisons of real output in manufacturing', Policy Planning and Research Working

Papers, WPS5, World Bank, Washington.

Machin, S. and Wadhwani, S. (1989), 'The effects of unions on organisational change, investment and employment: evidence from

WIRS', Centre for Labour Economics, Discussion paper no. 355, London School of Economics.

Maddison, A. (1989), The World Economy in the 20tb Century, OECD, Paris.

Maddison, A. (1991), Phases of Capitalist Development, Oxford University Press.

Matthews, R. and Feinstein, C. (1990),, The growth of output and productivity in the UK:The 1980s as a phase of the post-war

growth' National Institute Economic Review, August. London.

Metcalf, D. (1989), Water notes dry up.- The impact of the Donovan reform proposals and Thatcherism at work on labour

productivity in British manufacturing industry', British journal of Industrial Relations, vol. XXVII, no. 1, pp 1-32.

Muellbauer, J. (1991), Productivity and competitiveness' Oxford Review of Economic Policy, vol. 7, no. 3. p. 99-117.

Nordhaus, William D. (1972), The recent productivity slowdown', Brookings Papers on Economic Activity, no.3 p. 493-546.

OECD (1985),'Economic Surveys 1984/1985, Germany', OECD, Paris.

OFCD (1990),'Economic Surveys 1989/1990, Germany', OECD, Paris.

O'Mahony, M. (1991),'Comparative productivity in UK and German manufacturing industry', mimeo, NIESR.

Oulton, N. (1990), Labour productivity in UK manufacturing in the 1970s and 1980s', National Institute Economic Review,


Paige, D. and Bombach, G. (1959), A Comparison of National Output and Productivity, OEEC, Paris.

Prais, S.J. (1981a), Productivity and Industrial Structure, Cambridge University Press.

Prais, S.J. (1981b),'Vocational qualifications of the labour force in Britain and Germany', National Institute Economic Review,

no. 98, November.

Prais, S.J and Wagner K. (1988), Productivity and management: The training of foremen in Britain and Germany' National

Institute Economic Review, no. 123, February.

Rostas, L. (1948), Comparative Productivity in British and American Industry, Cambridge University Press.

Smith, A.D, Hitchens, D.M.W.N, and Davies, S.W. (1982), International Industrial Productivity, NIESR, Cambridge University

Press, Cambridge.

Steedman, H. and Wagner, K. (1987),, A second look at productivity, machinery and skills in Britain and Germany', National

Institute Economic Review, November.

Steedman, H. and Wagner, K. (1989), Productivity, machinery and skills: clothing manufacture in Britain and Germany',

National Institute Economic Review, May.

Summers, R. and Heston, A. (1988), A new set of international comparisons of real product and prices: estimates for 130

countries, 1950-85', The Review of Income and Wealth, series 34, no. 1, March.

Summers, R. and Heston, A. (1991),'The Penn world table (mark 5): an expanded set of international comparisons, 1950-1988',

The Quarterly journal of Economics, May, pp. 327-368.

Van Ark, B. (1990a), Comparative levels of labour productivity in Dutch and British manufacturing', National Institute

Economic Review, February, London.

Van Ark, B. (1990b), Manufacturing productivity levels in France and the United Kingdom', National Institute Economic

Review, August, London.

Van Ark, B. (1990c), 'Comparative levels of manufacturing labour productivity in postwar Europe', Oxford Bulletin of Economics

and Statistics, vol. 52, November, pp. 343-374.


Early research on measuring productivity levels across countries, for example, Rostas (1948) compared physical quantities of major goods produced (tonnes of steel, numbers of cars etc.). Since the Rostas study, physical indicators have became less representative of modern complex products so that an alternative method was developed, based on comparing the nominal value of total output per person converted to a common currency using the relative prices of manufactured goods in different countries. Studies of this type include Paige and Bombach (1959), Smith, Hitchens and Davies (1982), Maddison and van Ark (1988), van Ark (1990a) and (1990b) to name but a few. This article follows this tradition.

The first step in the calculation of relative productivity is to estimates the ratio of producer prices in the two countries. Producer price levels are not published in either country although indexes of price movements are available. Each country publishes data on both the value and the quantity sold of many products so that it is possible to calculate the average value of a particular group of manufactured products in both countries.

The calculations of unit value ratios were carried out for as many 3-digit industries as it was possible to find some matched products. The resulting unit values for product i in industry k and country i are denoted by [P.sup.j.sub.ik]. An average unit value ratio for all products of the industry, for which information is reported in the census, can then be derived using either of the following two formulas:


where Q refers to the physical quantity of matched product i and G and U refer to Germany and the UK, respectively. Note that the summation is over all matched products and not over the entire output of industry k. Formula (1) weights the prices of individual items using German quantity weights and (2) uses UK quantity weights; differences between (1) and (2) will reflect differences in the importance of individual matched items in the two countries.

In most industries matching involves only a subset of all goods produced. To calculate relative levels of nominal output, and hence labour productivity, it is necessary to convert the entire output of each industry to a common currency. Therefore, we employ the assumption that the relative prices for matched products is representative of the relative prices of all products in the industry. Formulas (1) and (2) are therefore used to convert total industry output to a common currency.

The use of unit values can be criticised on three fronts. The first is that they may not be equivalent to relative prices since the matched items may not be exact substitutes for each other. Some judgment was exercised here but, in general, we erred on the conservative side, only matching items whose product description was very similar. As mentioned in the main text, this procedure does not adjust for differences in quality.

The second problem is that the matched items may not be representative of the total output of the industry. The percentage of output which can be matched depends on the availability of both the quantity and value data for specific items. The quantity sold of many items are not disclosed either due to reasons of confidentiality or because there are too many varieties making it difficult to define and hence measure the physical unit (for example, machine tools).

Finally, the UVRs measure relative prices of gross sales and are then applied as conversion factors to gross value added. Thus we are assuming that the relative prices of gross output and gross value added are equal, that is, that the relative prices of material inputs are the same across countries. The use of this single deflation method is usually justified by arguing that countries face similar prices for raw materials such as metals, oil and so on since these products are generally traded at world prices. Price levels for material inputs are not generally available so single deflation is used throughout this study.

It proved possible to derive unit value ratios for 250 manufactured products. Table A-1 shows the percentage of sector sales which were accounted for by matched items and the number of individual items matched. It also shows the geometric mean of the unit value ratios calculated using formulas 1) and (2). The percentage of sales matched varied considerably across sectors as did the estimated unit value ratios.

Table A-1 shows that it was not possible to match any items in both mechanical and instrument engineering and only a small percentage of sales were matched in electrical engineering, all of which were consumer goods. Therefore, an alternative conversion factor had to be used for these industry branches. Using the average UVR for all manufactured products or that for vehicles was deemed inappropriate given the large variation of UVRs across the industry branches for which reasonable matches were possible. instead, use was made of the information provided by the 1985 purchasing power parities (PPPs). The ratio of the engineering PPPs to that for total manufacturing was then applied to the average manufacturing UVR to yield a 'relative UVR' for these branches.

In the case of electrical engineering, the relative UVR was used as an estimate of relative prices for capital goods but the UVR based on matched items was used for consumer goods. The above procedure for estimating UVRs in engineering assumes that differences between UVRs and PPPs are proportionately the same across industries.

Measurement of the relative levels of output per worker require data on the value of output and employment on a consistent definitional basis in the two countries. This is most readily achieved using returns from the censuses of production; these have the advantage that both output and employment are derived from returns from the same reporting units.

This paper measures output as gross value added which is defined as gross output minus the cost of purchased materials, industrial and non-industrial services. Previous studies, for example, Smith et al (1 982) used net output, (gross value added plus the cost of non-industrial services) since gross value added was not included in the UK census until 1973. The use of gross value added lowers the Germany [UK productivity ratio relative to its level using net output, since German manufacturing industry appears to purchase a greater proportion of non-manufactured inputs. The British output data were adjusted for stock appreciation, estimates of which were taken from the 1991 Blue Book. The British data were also adjusted to exclude firms employing less than twenty workers since these firms are excluded from the German source. This was achieved using unpublished data from the census of production which we acquired from the CSO.

Data on average annual hours per worker were readily available for Germany but not for the UK. Estimates, by industry, for the UK were derived by updating the estimates for 1984 used in van Ark (1990a), described in Annex C of that paper. Average hours worked per week were estimated by weighting average hours for various types of worker (male, female, full-time, part-time, manual, non-manual) by their share in employment and included an allowance for short-time working. Average number of weeks worked per year included allowances for annual leave, public holidays, sickness and maternity leave, and days lost through strikes.


This research was financed by a grant from the Leverhulme Trust to which I owe thanks. I am particularly grateful to Bart van Ark for data and comments throughout the research and to Nick Oulton and Sig Prais. I would like to thank Tony Smith for making available the worksheets from the previous NIESR study on German and UK productivity. Helpful comments were also received from participants at seminars at NIESR and the Statistisches Bundesamt, Wiesbaden, Germany, and from a referee, Steve Broadberry, Steve Davies, Angus Maddison, Sheila Page and Donald Roy.

1 The productivity comparisons do not cover what was in 1987 the German Democratic Republic so that in the text and tables, Germany' refers only to Western Germany'.

2 See, for example, Hitchens, Birnie and Wagner (1990), Hooper and Larin (1989) and van Ark (1990c).

3 These were the Report of the Census of Production, 1987', CSO, for the UK and `Kostenstruktur im Bergbau und Verarbeitendes Gewerbe, 1987', Statistisches Bundesamt, for Germany.

4 It should be noted that the BLS series are primarily concerned with cross country estimation of growth rates in output and labour input and the BLS specifically warns against the use of their figures for levels comparisons.

5 Van Ark (1990c) also presents an estimate of about a 40 per cent productivity gap between Germany and the UK for the late 1980s, based on an update of the 1968 estimate of Smith et al, using time series on real output and employment in both countries. Subsequently, the `Statistiches Bundesamt' has revised considerably their output and employment series. In addition, the time series for UK average weekly hours worked in this paper differs from van Ark's in that it more accurately measures changes due to increased paid holidays.

6 These authors did not include an adjustment for average annual hours worked.

7 This analysis is limited by the fact that it is carried out for only 14 industry branches and so does not capture the shift in employment within these industries. Also, the calculation was carried out for output per worker since data on average hours worked by industry were not available in 1968.

8 As shown in van Ark (1990b) Annex C.

9 Gross capital stock were estimated at the industry branch level and then aggregated up to total manufacturing. This increases the accuracy of the estimates since it adjusts for the fact that asset lives may vary across countries due to different industry shares in total investment.

10 Hellwig and Neumann (1987) suggest that the need for firms to 'socially justify' dismissals increased throughout the 1970s and early 1980s.
COPYRIGHT 1992 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 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:includes appendix
Author:O'Mahony, Mary
Publication:National Institute Economic Review
Date:Feb 1, 1992
Previous Article:The world economy.
Next Article:Housing and the wider economy in the short and long run.

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