A Comparative Labour Productivity Benchmark for
Germany and Britain in 1951
By: N.E.S. Bos S1529889 Supervisors: Herman de Jong Tamas Vonyό August 2010 Abstract
My thesis provides a comparative labour productivity benchmark of Anglo-German comparative productivity for the manufacturing sector in the year 1951, at the beginning of the Golden Age of economic growth. This study applies the industry of origin approach, which relies on information from production censuses to derive unit value ratios, which can be used to convert two currencies. The main result is a substantial labour-productivity gap in the manufacturing sector. Output per hour in the United Kingdom was significantly higher than in Germany. This gap is partly explained by a substantial difference in capital intensity and the skilled-labour ratio of the workforce.
Table of contents:
1. Introduction 5
1.1. Previous benchmark comparisons 6
1.2. Literature on the Golden Age 8
1.2.1. British decline 8
1.2.2. The impact of the Second World War 9
1.2.3. Catching up and convergence 11
1.2.4. The role of institutions 11
1.3. Objective 12
2. Data and Methodology 13
2.1. Data 13
2.2. The year of comparison 15
2.3. Hours worked 16 2.4. Methodology 19 2.4.1. Deflation methods 22 2.5. Matching of products 22 2.6. Labour-productivity estimates 23 3. Analysis 28
3.1. Shift share analysis 29
3.2. Regression analysis 32
3.2.1. Capital intensity 32
3.2.2. The scale of production 34
3.2.3. Skill intensity of the labour force 35
3.2.4. Wages and salaries 37
3.2.5. Results 38
4. Conclusion 42
List of Tables:
Table 1: Annual hours worked in the manufacturing sector in Germany 18 and Britain 1950
Table 2: Coverage ratios in the 12 main industries 24
Table 3: Gross output PPPs Germany and the United Kingdom (1951) 26
Table 4: Static shift-share analysis - Germany and the United Kingdom (1951) 30
Table 5: Dynamic shift-share analysis - Germany and the United Kingdom (1951) 31
Table 6: Regression results of OLS estimation and robust regression estimation 38
Table A1: Previous benchmarks comparisons 48
Table A2: Reclassification of manufacturing branches in Germany 50 and the United Kingdom
Table A3: Conversion factors for imperial units and energy 53
Table A4: Purchasing Power Parities Germany and the United Kingdom (1951) 54
Table A5: Purchasing Power Parities Germany and the United Kingdom (1951) 55
weighted by gross output
Table A6: Labour-productivity per branch in manufacturing in Germany and the 56 United Kingdom (1951)
Table A7: Shares of industries (measured by hours worked) 57
Table A8: Education attainments of population aged 15-64 in 1950 58
List of figures:
Figure 1: Per Capita GDP levels in the period 1937-1962 (1990 International 16 Geary-Khamis dollars)
Figure 2: Annual hours worked in Germany and Britain 1870-2000 19
Figure 3: Robustness check: Maximum coefficient of variation 25
Figure 4: Comparative labour productivity per branch in manufacturing in 27 Germany and the United Kingdom (1951)
Figure 5: Industry shares in total manufacturing (1951) 29
Figure 6: Energy usage in the manufacturing sector in Germany (1951) 33
Figure 7: Energy usage in the manufacturing sector in the 33 United Kingdom (1951)
Figure 8: Female labour participation in the manufacturing sector in 37 Germany and the United Kingdom (1951)
Figure 9: Two-way scatter – labour-productivity ratio and wages 37 and salaries ratio (1951)
Figure 10: Leverage versus squared residuals plot 39
Figure A1: Coal usage in manufacturing industries in Germany and 60 the United Kingdom (1951)
Figure A2: Coke usage in manufacturing industries in Germany and 60 the United Kingdom (1951)
Figure A3: Electricity usage in manufacturing industries in Germany and 61 the United Kingdom (1951)
1. Introduction
Since the mid-1980s we have witnessed a major resurge in empirical and theoretical research on economic growth. This also renewed interest in the high growth period 1950 to 1973, known as the Golden Age1. During this period, output per hour worked in the manufacturing sector grew 6.62 percent per annum in Germany and 4.69 percent per annum in the United Kingdom2. These growth rates were much larger than the growth rates before and after this period. Although a large body of literature is available on theories of economic growth, there is a need for better historical evidence to analyze the comparative performance of manufacturing in the early 1950s. Most of the existing international comparisons of productivity levels concern the United States and the United Kingdom. Germany has been largely neglected, although it was one of the mayor players in the world economy in the twentieth century. The objective of my thesis is to provide a clear, consistent benchmark for Anglo-German labour productivity for the period directly after the Second World War.
Productivity comparisons are closely related to the core of economics and have a long history. Already in 1690 Sir William Petty published a comparison of wealth between France, England and Holland for 16753. This was followed by a comparison for 1696 by Gregory King who used income, expenditure and production information from national accounts4.
International comparisons of levels are mostly made from either the expenditure side or the production side of the economy. The expenditure approach concentrates on categories of private consumption, government consumption and capital formation5. The first attempt to compare real output and productivity from the production side of the economy was made by Rostas, who compared productivity in the United States and United Kingdom for the second half of the 1930s6. He applied the so-called “industry of origin” approach. This approach aims at comparing levels of output by industry, rather than comparing expenditure categories. The existing pre-1945 benchmarks are mainly based on a direct comparison of physical output per worker. The post-1945 benchmarks are obtained using the methodology outlaid by Paige and Bombach7. The major novelty of the Paige and Bombach study was the use of data on net output available in census data. Net output is defined as gross output minus the costs of materials and fuel used and the amount paid to industrial services. The censuses provide data on both the quantity of sales in physical units and the value of these sales. Hence, the factor-gate price or unit
1 Crafts 1995.
2 O’ Mahony 1999, p. 14. 3 Sir William Petty 1690. 4
King 1696.
5 Van Ark 1996. 6 Rostas 1948.
values of a product can be obtained through dividing the value of sales by quantity. Paige and Bombach converted values in different currencies by using derived unit values from national production census data. Paige and Bombach were the first to apply this approach in a comparison of the United States and the United Kingdom. Both the direct comparison method and the method of Paige and Bombach offer a solution for the problems that plagued earlier research, which used the exchange rate to convert values of output in different currencies 8.
The problem with the use of exchange rates is that the exchange rate might not reflect the correct overall purchasing power parity (PPP) between the currencies under consideration. Capital movements can play a major role in the determination of exchange rate levels and can lead to fluctuations. Moreover, it fails to take into account that the purchasing power of a currency will normally differ for different products. For the exchange rate to be a valid deflator, the internal price level and structure of a country must coincide with world market conditions.
1.1 previous benchmark comparisons
Several studies compared labour productivity between Britain and other countries at different benchmarks since 19079. Table A1 in the Appendix shows an overview of the main results of the studies on Anglo-German productivity, which will be discussed below.
Rostas made a comparison for Germany, the United States and the United Kingdom for the interwar period10. Rostas used two methods of comparison: one based on physical output per head and one based on the value of this output. The physical indicators facilitate a direct comparison of labour productivity between countries in an industry. Rostas found that the result of the two methods were not significantly different. Germany was superior in coal-mining, coke manufacturing, cotton spinning and, rayon and silk. Whereas Britain was superior in clay and timber trades, paper trades and food, drink and tobacco trades.
Broadberry and Fremdling provide new estimates of physical output per worker for 1935 and estimates over time from 1907 to 193711. They found that the most important proximate cause of the productivity differential is relative plant size. Britain was less productive in small plant metals and engineering sectors. Moreover, collusive behavior and cartelization were found to be important in explaining the failure of Britain and Germany to respond to the challenge of much higher productivity levels in the United States. Germany was found to be relatively most productive in the heavy industries, such as coal mining, steelworks and iron foundries. Whereas
9 See Rostas 1948, Flux 1933, Paige and Bombach 1959 and, Frankel 1955. 10 Rostas 1943.
Britain was relatively most productive in the lighter industries such as food, drinks and tobacco and cotton weaving.
Fremdling, de Jong and Timmer provide a new industry of origin study of Anglo-German manufacturing output and productivity levels by industry for 1935-1936 based on so-called double deflation12 , where both gross output and intermediate inputs are deflated. They also adjusted for the production in protected industries. The aggregate result is close to the finding of Broadberry and Fremdling, who used a quantity approach and a smaller set of industries13. The rank order of comparative productivity levels among industrial branches did not change either. However, the productivity difference between Germany and Britain became more pronounced in most cases when the double deflation approach was applied.
Smith, Hitchens and Davies made an Anglo-German comparison for 1967-1968. They used in some instances market prices instead of factor prices, and in some cases physical indicators14. They covered 39 percent of the German and 37 percent of British value added, and showed a considerable Anlgo-German productivity gap. Van Ark used the comparison of Smith, Hitchens and Davies and extrapolated it on the basis of time series on output and employment15. For 1950 he found the value added per person-hour in German manufacturing to be 74 percent of the British level16. O’Mahony made an Anglo-German comparison for 1987 based on 236 matches, which compares to 22 percent of gross output17. The pattern of advantages she finds is comparable to the findings of Rostas. However, when time-series of different origin are projected from a benchmark-year into distant periods the validity of the comparisons depends on how stable the basket on which the Purchasing Power Party converters have been established remains over time. Relative prices would usually change over time, rendering the base year weight obsolete. Therefore, the economic meaning of a comparison based upon remote PPPs is entirely questionable, and the errors obtained in conversion might be larger than the bias when exchange rates are used18.
After the 1980s many productivity studies which focused on these countries exist. However, these do not focus on the immediate post-war period and are therefore not of a major interest for this thesis. The studies mentioned above rely on different methods and datasets, and, therefore, a direct comparison of their results is not feasible. The productivity advantages are not
12 Fremdling, de Jong and, Timmer 2007. 13 Broadberry and Fremdling 1990. 14 Smith, Hitschens and Davies 1982. 15 van Ark 1990.
16
ibid, 345. Note: The output figures refer to “net output at market prices” which includes the costs of service inputs and net indirect taxes.
the same in the above mentioned studies19. The lack of consistency in methodology and classification of industries prohibits a comparison of benchmarks over time. In this thesis I will provide a labour-productivity benchmark that is comparable to the 1935-1936 benchmark of Fremdling, de Jong and Timmer facilitating a direct comparison between two periods.
1.2 Theories on the Golden Age
There are a few economic historical and theoretical issues important in the light of this thesis. The labour-productivity benchmark I provide can be used to study these theories in more detail. I will start with a discussion on British decline. Then I will discuss reconstruction dynamics, catch up and convergence, and the effect of institutions on the golden growth period.
1.2.1 British Decline
In the first half of the nineteenth century, Britain was the leading manufacturing country, and was seen as the workshop of the world20. However, that leadership passed to the United States in the twentieth century. Britain ranked second in real income per person in Europe in 1950, but fell to the tenth position by 197921. During the 1950s Britain failed to reduce the transatlantic productivity gap and was rapidly caught up and overtaken by major European rivals22. The Anglo-German productivity gap in manufacturing widened between 1952 and 1959, and stabilized from 1959 to 197323.
During the Golden Age, British economic growth appears to have been exceptionally disappointing, when placed in a comparative international context24. Literature on British industrial performance during the Golden Age of economic growth has emphasized competitive weakness and deindustrialization. The share of world manufacturing fell dramatically during this period. Crafts reports that the share fell from 20.9 percent in 1937 to 16.5 percent in 1960 and only 9.1 percent in 197925. Literature on British decline usually assumes that the United States and Germany were able to overtake Britain because of the emerging lead in the manufacturing sector26. Especially Crafts strongly emphasized the Manufacturing Failure Hypothesis, which states that the relative decline of Great Britain’s prosperity is caused mainly by low productivity in the manufacturing sector and failure in large scale production. Broadberry however, argues that
19 For example, in the Fremdling, de Jong and Timmer study the advantage of leather trades is for the UK,
whereas in the Smith, Hitchens and Davies study the advantage is for Germany.
20 Royle 1987, p. 36. 21 Maddison 1992.
22 Broadberry and O’Mahony 2004. 23
Broadberry 1997, p. 49-50.
24 Bean and Crafts 1996, p. 142. 25 Crafts 1993, p. 20.
the catching up of the United States and Germany over the United Kingdom was mainly a result of sectoral shifts outside the manufacturing sector. According to Broadberry the United Kingdom was very efficient in the service sector, giving scope for Germany and the United States to catch up.
Broadberry and Crafts believe that the cause of the British productivity debacle in the 1970s was the culmination of the longer run problems in manufacturing that centered on weak incentive structures and institutional arrangement in the “postwar settlement” 27.
Not all scholars agree on the idea that British decline is caused by failure in the manufacturing sectors. Booth for example reasons that manufacturing performance of Britain could have been higher during the Golden Age, but that other sectors also are of great importance for the total economic performance. Booth also shows that Britain did not de-industrialize during the economic boom, since even in 1973, manufacturing contributed to a larger share of GDP in Britain than in most other countries28.
1.2.2 The Impact of the Second World War
According to the reconstruction thesis of Ferenc Jánossy, war-shattered economies automatically recover to their long-run productive potential29. According to this thesis, the process of post-war reconstruction generates extraordinary growth potential. The Reconstruction Thesis assumes that the destruction of physical capital and underinvestment in new equipment and plant during the war, induces a mismatch in the factors of production (labour and capital) after the war. The low initial capital-labour ratio leads to high returns on capital, inducing high levels of investment and thus securing rapid productivity growth. The Jánossy thesis has mainly been applied in the German literature30. Abelshauser argues that after the Second World War, there was a large discrepancy between actual output and the productive potential determined by the country’s pre-1914 growth path31. Abelshauser and Dumke argue that the larger the drop in output that a country suffered between 1938 and 1950, the faster it will grow subsequently32. Dumke argued that greater wartime destruction generated faster postwar growth and provided evidence for this proposition in cross country growth regressions for OECD countries33. Dumke also stresses the similarities between the reconstruction thesis and the productivity gap hypothesis. Abramovitz however, asserted that the destruction of physical capital during the war was less
27 Broadberry and Crafts 1992. 28 Booth 2003, p. 7.
29 Jánossy 1969. 30
See for example Abelshauser 1975 and 1983; and Borchardt 1991.
31 Abelshauser 1983, p.92. 32 Idem 1981.
important than the maintenance of what he calls “social capability” of growth34. This is in line with Jánossy’s human capital determined long-run growth potential.
As stressed by Vonyό there remains more disagreement than consensus about the explanatory power of the Janossy model against more mainstream growth models35. Vonyό found that for the core western industrialised nations, the rapidity and variety of economic growth during the 1950s and 1960s can mostly be explained by post-war reconstruction. Crafts and Mills also found evidence for the reconstruction thesis in ten European countries, although they did not found evidence for the longer run dynamics of reconstruction36.
The material damage in West Germany was of a small magnitude according to the United States Strategic Bombing Survey (USSBS). The survey stressed that the Allies never attempted to destroy the German economy, but rather to stop it from operating by damaging key points37. Beginning mid 1944 Allied Strategic Bombings targeted electric power, synthetic fuels generation, and railroad networks, in an attempt to disrupt the supply chain instead of destroying productive capacity38. Only 17.4 percent of industrial fixed assets on the territory of the later West Germany was destroyed as a consequence of aerial bombardment and ground fighting, and a mere 6.5 percent of all machinery and equipment suffered significant damage39.
Although human causalities during the war were enormous, the labour force in West Germany was substantially larger at the start of the 1950s than before the war, due to the large inflow of refugees from the provinces ceded to Poland and USSR as well as from Eastern Europe40. These minorities were expelled from their traditional settlements areas in East and Central Europe in accordance with the Potsdam agreement. Overall West Germany’s population increased by 9 percent from 1946 to 195041.
The literature on reconstruction in Germany stresses that the reconstruction dynamics were not restricted to the period directly after the war, but manifested itself until the mid 1960s. In order to reinvestigate the importance of post-war recovery in German supergrowth vis-à-vis Britain, we need a consistent benchmark for the two countries at the beginning of this growth period.
34 Abramovitz 1986. 35 Vonyό 2008.
36 Crafts and Mills 1996. 37 USSBS 1945, p. 37. 38
Birkenfeld 1964.
39 The 1943-1945 level compared to 1943. Krengel 1958 cited by Abelshauser 2004, p. 68. 40 Steinberg 1991, Eichengreen and Ritschl 2008.
1.2.3 Catching up and convergence
The convergence and catch-up view tries to explain the rapid growth of the German economy after the Second World War by emphasizing structural change and productivity growth42. The scope for catch up and reconstruction was clearly less for Britain than for Germany. German GDP per man hour (in the total economy) was never as much as 75 percent of British GDP per man hour at any point before 1950 but converged to British levels in the 1960s43.
Part of the faster economic growth in Germany compared to the United Kingdom was due to the more rapid sectoral shift in the German economy44. In 1950, 24 percent of Germany’s labour force was employed in the agricultural sector compared to 5 percent in the United Kingdom45. Due to the low share of labour force in agriculture at the start of the Golden Age, there was less scope for sectoral shifts in Britain. Several studies pointed out that the United Kingdom had a higher initial level of aggregate labour-productivity, and therefore lacked the potential for rapid catch up46. According to Kaldor’s law, the growth of an economy ultimately depends on the growth of the industrial production, and the percentage this represents in total output47. This indicates how important it is to understand growth in the manufacturing sector.
To do in depth research on the importance of catch up and convergence on the enormous growth during the Golden Age, a detailed benchmark at the beginning of the Golden Age period is needed.
1.2.4 The role of Institutions
According to Olson long-standing distributional coalitions were dissolved by the war, freeing Germany to enjoy a sustained acceleration in total factor productivity growth48. British industrial relations were not reformed as a result of the war, traditions of voluntarism and multi-unionism continued49.Elbaum and Lazonick perceive the decline of the British economy to be the result of rigid persistence of economic and social institutions from the nineteenth century50. Whereas Germany, the United States and Japan experienced successful economic development based on mass production and corporate forms of managerial coordination, Britain lagged behind. Eichengreen provides a full account of the episode of “coordinated capitalism” in which institutions, neo-corporatist bargain between employers, trade unions, and government were an
42 Abramovitz 1986.
43 Eichengreen, and Ritschl 2009, p.2. 44 Temin 2002, p.9.
45 Broadberry 1997, p.252.
46 See e.g. Crafts 1995, p.249 and Feinstein 1994. 47
Kaldor 1967.
48 Olson 1982. 49 Crafts 1995.
important factor in catching up51. Eichengreen argued that a “social contract” between labour and capital established a dynamic game to secure wage restraints, in exchange for commitment to reinvest profits52.
1.3 Objective
My thesis will provide a comparative labour-productivity benchmark for Germany and Britain closely after the Second World War at the beginning of the Golden Age of economic growth. I will focus on West Germany, to which I will refer hereafter as Germany. Although there are comparisons for the interwar period, there is no benchmark for these two major players in the world economy at the beginning of the 1950s. The benchmark is needed to do research into the above mentioned theories. Detailed insights into the productivity differences between Germany and Britain at the start of this high growth period, which is currently not completely understood, will be obtained. As explained above, there remains much uncertainty concerning the supergrowth period. With a good benchmark for the early 1950s, it becomes possible to investigate the impact of the Second World War and how much of growth can be explained by rebounding from the war-induced shock. British decline, the reconstruction dynamics, and catch-up and convergence within the manufacturing sector can be better understood once a labour-productivity benchmark at the beginning of the Golden Age is available.
Although the aim of my thesis is not to explain the observed productivity gaps, the industry of origin approach provides useful tools to analyze the impact of compositional differences in manufacturing activities on productivity levels. With use of shift-share analysis, I will attempt to explain the productivity differences existing in 1951. I will use regression analysis to analyze the impact of scale effects, skill intensity in the labour force, and capital in explaining the Anglo-German productivity differences.
The rest of this thesis is constructed as follows. The next Chapter explains the data and methodology used in this study, and presents the labour-productivity benchmark. The third Chapter presents the analytical investigation of the labour-productivity gap. The final Chapter consists of the conclusion.
2. Data and Methodology:
In the first part of this Chapter, I will provide a thorough description of my data-sources, and the advantages and disadvantages of these data. In the second part of this Chapter, I will explain the methodology used to establish labour-productivity benchmarks.
2.1 Data
The data for the benchmark comparison came from the official production censuses of Britain and Germany. Production censuses provide the most reliable data for productivity comparisons, since one source is used to obtain information on gross output, value added, and employment, guaranteeing internal consistency. For the United Kingdom, I used the The Report
on the Census of Production for 1951, published by the Business Statistics Office of the Board of
Trade53. The area covered by the British census does not include Northern Ireland, which the censuses of 1935 did. For Germany I used the annual industry statistics published by the Federal Statistical Bureau54.
The German statistics provides information on five core industry groups: Mining, raw material and producer goods industries, investment industries, consumption goods industries and, food, drinks and tobacco industries. These industries are broken down into 44 industry branches, which have been split up in sub-branches in some cases. The British industrial sector encompasses 24 industries subdivided into 148 industry branches. I have reclassified the industries from both countries into 12 branches: Textile trades, leather trades, clothing trades, iron and steel trades, engineering and vehicles, non-ferrous metal trades, food, drink and tobacco trades, chemical and allied trades, timber trades, paper trades, and miscellaneous trades. I thereby adhere to the classification of Fremdling, de Jong and Timmer and de Jong and Woltjer, facilitating a direct comparison of the results55. I will also present my results on a more disaggregate level. Table A2 in the Appendix shows the classification of industries.
In the British census, establishments were classified to trades according to the nature of their output. An establishment engaged in multiple activities, e.g. a firm engaged in machines production and casting, was classified to a trade if its production of principal products of that trade accounted for a greater proportion of the value of its output than did its production of
53 Board of Trade 1954.
54 Die Industrie der Bundesrepublik Deutschland” published by the Statistisches Bundesamt 1956. The
issue Employment and Sales, Fuel and Energy Supply 1951-1955 contains information on employment, while the issue The Industrial Production 1951-1955, contains detailed information on production and value addedIn German: “Beschäftigung und Umsatz, Brennstoff- und Energieversorgung 1951 bis 1955” and "Die Industrielle Produktion 1951-1955.
principal products of any other trade. Offices warehouses, laboratories and other ancillary places of business, which were separated apart from the producing work, were not regarded as separate establishments, and the persons employed were included on the return for the works. If firms with more than one establishment were unable to make separate returns for each establishment, they were generally allowed to make one return covering all establishments in one trade.
In the German census firms active in multiple industries, were placed in the industry group where the core of the business was, as measured by the number of employees engaged in the production. This method of classification differs from the British method, where the value of output was used to find the core business. However, it seems reasonable to expect that these methods will not deviate too much, since a larger value in output and more employees are highly correlated56.
One possible problem with using censuses is that they often omit data of smaller firms. If the omission is larger in one country than another comparisons become inconsistent57. In Britain proprietors employing an average of under ten people were not required to report detailed returns. However, small firms were required to provide information on the average number of male and female workers and the nature of their business. In trades in which the output of small firms was thought to be a relatively high proportion of the total output, small firms were required to comply a simplified form. In the German census the same rule applied, information is provided only for firms that employed at least ten persons. The German census provides no information when there are less than three firms operating in an industry for confidentiality reasons. Thus, the comparative benchmark will not include information on handcraft establishments, but since the omission is the same in both countries, the benchmark will be consistent.
Comparative productivity studies inevitable raise issues of quality differences. Even the most detailed product description will not completely pick up the differences in quality. There is no real solution to the quality problem in a cross-country comparison. In a time series analysis one can expect the quality of products to increase over time, but in a cross-country comparison there is no method to know in which products/sectors there is a quality advantage and in which there is a quality disadvantage58. However, in 1951 this issue was not as severe as it is today59. Broadberry and Crafts show that the comparative productivity picture they obtain when using net output converted using relative unit value ratios is highly similar to the comparison of Frankel who uses physical output for a comparison for the United States and the United Kingdom in
56 For Germany the correlation between the net value and number of employees is 0.834 (p-value <0.001)
and for the United Kingdom this correlation is 0.971 (p-value < 0.001).
57 Fremdling, de Jong, and Timmer 2007. 58 Van Ark, 1993, p.36.
194860. Moreover, quality differences are more important in consumer durables and investment goods than in basic goods, such as steel, cement, paper, wood etc. and especially the latter make up a large part of manufacturing output61.
Another issue is that there are virtually no cases in which products are truly homogenous. Even for simple products as marmalade, much quality differences can exist. One problem is that the British census provides information on a more disaggregate level than the German census. For example, the British census provides information on dressmaking of different materials (wool, silk rayon, nylon, and other textiles materials) whereas the German census only provides one entry for women’s and girls’ dresses. Where possible I added the British products into groups that I could compare with the German product. In some cases this adjustment was easy, since the good description was clear. I could for example just add the dresses in one group, to obtain a product comparable to the German product. However, in other cases the aggregation of the British products was less straightforward.
In cases where I added multiple products to form a product group that could be compared to a product from the German census I first carefully examined the descriptions of the products. Then, after adding the British and sometimes the German products to form a comparable product group I checked the unit value ratios of all products. In some cases, where it was highly uncertain whether a product with an extreme UVR, and a low share, did belong to a product group, I did not use that product in the benchmark.
Furthermore, the degree of differentiation of both intermediate and end products have a good deal to do with differences in production methods and per worker output levels. The degree to which industries are vertically integrated differs between the countries.
2.2 The year of comparison
As mentioned by Rostas, a comparison is only relevant if the years compared between the two countries are reasonably comparable as to the rate of unemployment and capacity utilization62. For both the United Kingdom and Germany a production census is available for 1951.
The United Kingdom was experiencing low unemployment, of 2.5 percent in the beginning of the 1950s whereas Germany’s unemployment rate was rather high, with 8.2 percent in the beginning of the 1950s63. However, this high unemployment is also caused by the large inflow of refugees and thus not an indication of failing labour market institutions or cyclical factors. Although the capital utilization and unemployment figures are clearly not the same in the
60
Broadberry and Crafts 1990, p. 376-377.
61 Van Ark 1993. 62 Rostas 1948.
United Kingdom and Germany in 1951, this is not a problem for this study. The literature on the Golden Growth period tries to emphasize the differences between Germany and the United Kingdom at the starting point of this period.
Figure 1 present per capita GDP levels of Germany and the United Kingdom in the period 1937-1962.
Figure 1: Per Capita GDP levels in the period 1937-1962 (1990 International Geary-Khamis dollars)
0 1.000 2.000 3.000 4.000 5.000 6.000 7.000 8.000 9.000 10.000 1937 1939 1941 1943 1945 1947 1949 1951 1953 1955 1957 1959 1961 G D P p e r c a p it a United Kingdom Germany Source: Maddison (1992)
There is a sharp decline in Germany’s GDP during the last years of the Second World War and a rapid recovery afterwards. In Britain, the dislocation from the growth trajectory is much milder, and there is no sharp recovery. However, the growth performance in 1950 seems not to be driven by the remaining distance from the steady state point in 1941, but rather from the dislocated position around 1945. In this sense, it would not be neoclassical growth that explains the sharp recovery but rather reconstruction dynamics64. In order to further study the Wirstchaftwunder and the growth dynamics of the Golden Age, a satisfactory comparison at the starting point of this period is needed. Thus, for my thesis the year 1951 is highly suitable as year of comparison.
2.3 Hours worked
The production censuses provide only information on paid employees. In some industries, and in particular in the smaller establishments and in the agricultural sector, we might expect that family workers play an important role. However, in this study I only investigate lager (10
employees or more), industrial establishments, thus the problem of uncounted family workers will be relatively minor.
Usually labour-productivity comparisons are provided in terms of man-years. However, due to differences in the length of the workweek and vacation days this concept may vary between countries. Therefore, a comparison based on man-hours might be preferred. The difference between productivity based on man-years and man-hours can be quite large65. However, Kravis warns that these man-hour figures are apt to be subject to greater error66. Moreover, he argues that using man-years has the advantage of indicating the industry's absorption of the nation's supply of workers.
The data of Mary O’Mahony provides information on the hours worked in Britain and Germany67. Table 1 provides data on the annual hours of work and number of vacation/holiday days per worker. As can be seen from Table 1 the workweek in Germany was considerably longer than the workweek in Britain. Figure 2 below shows the annual hours worked in Germany and Britain for the period 1870 to 2000. Clearly, Germany had significantly more hours worked per employee per year until 1929. In advance of the Second World War the German workweek was shorter than the British workweek. In 1948 the workweek was 42 hours in Germany, but after repair investments it returned to the normal level of 48,2 hours a week in 1950 68.
Part of the longer working hours in Germany is explained by the fact that Germany has very high “extra hours” in the beginning of the 1950s. Schudlich reports that these additional hours worked were more than two hours per week per person69.
These significant differences in the length of workweek and annual hours worked will have an effect on the labour-productivity comparison. If we fail to take into account Germany’s longer workweeks, we will underestimate the labour productivity Britain.
There are also pronounced differences over industries in the annual hours worked. The difference between the industry with the lowest number of hours worked -leather and footwear industry- versus the industry with the highest annual hours worked - non-metallic mineral products industry- is respectively 314 and 373 hours per annum for Britain and Germany.
65 Frankel (1955) reports a physical output per worker US/UK ratio of respectively 4.03, 0.95, 1.76 and
4.17 for wool-yarn, cured fish, rubber tires and pig iron. Whereas the respective physical output per man-hours ration US/GB are 4.53, 1.2, 2.03 and 4.91 for wool-yarn, cured fish, rubber tires and pig Iron.
66
Kravis 1976.
67 O’Mahony 1999.
Table 1: Annual hours worked in the manufacturing sector in Germany and Britain 1950
Industry Sector Average annual hours per person engaged,
manufacturing
United Kingdom Germany Ratio (GE/UK)
Chemicals and allied products Total 2042 2325 1.14
Chemicals 2027 2345 1.16
Rubber & plastic 2075 2257 1.09
Basic metals & fabricated metal
products Total 2095 2356 1.12
Basic metals 2137 2374 1.11
Metal products 2032 2313 1.14
Motor vehicles and engineering Total 2079 2323 1.12
Office & mech. Engineering 2127 2420 1.14
Mechanical engineering 2130 2433 1.14
Office machinery 2077 2138 1.03
Electrical engineering 2007 2214 1.10
Motor vehicles 2062 2265 1.10
Other transport equipment 2117 2378 1.12
Instrument engineering 1996 2321 1.16
Textiles, clothing and leather Total 1941 2145 1.11
Textiles 1994 2164 1.09
Leather, footwear& clothing 1880 2124 1.13
Food drink and tobacco Total 1998 2422 1.21
Other manufactures Total 1970 2432 1.23
Non-metallic mineral products 2194 2497 1.14
Wood & furniture 2099 2462 1.17
Paper & printing 2055 2368 1.15
Miscellaneous Manufacturing 1962 2246 1.14
Total manufacturing 2070 2327 1.14
United Kingdom Germany
Number of vacation and holiday
days* 24 29
Source: O’Mahony 1999, appendix table C p, 96-103.
Data is for manual workers only, data for Britain is based on series available in British Labour Statistics: Historical Abstract
1886-1968 and British Labour Statisics yearbooks. Data for Germany is based on unpublished series made available by the DIW.
Figure 2: Annual hours worked in Germany and Britain (1870-2000) 0 500 1000 1500 2000 2500 3000 3500 1870 1880 1890 1900 1913 1929 1938 1950 1960 1973 1980 1990 2000 year h o u rs p e r y e a r Germany Britain
Data source: Huberman and Minns 2005
2.4 Methodology
This thesis is based on the industry of origin approach. The industry of origin approach relies on unit values to convert values into a common currency. As explained by van Ark, unit value ratios are the most appropriate indicator for price comparisons in manufacturing. Purchasing Power Parities (PPPs), as calculated by the International Comparison Program, are designed for expenditure comparisons and will lead to biases when used in a productivity comparison. PPPs include relative transport and distribution margins, and foreign prices, and are usually expressed at market prices. However, market prices include value added tax and excise duties, which are difficult to take out. Another advantage of using the unit value method is that the production census provides data on sectors that produce mainly intermediate inputs. For example, pig iron, basic chemicals and paper pulp are not sold for final consumption, but only used as intermediate inputs in the production of other goods. If we would use expenditure prices to obtain a benchmark, these sectors would not be covered. In this thesis the industry-of-origin approach is applied and the necessary unit values are obtained by diving ex factory sales value (o) by the corresponding quantities (q) for each industry i. The unit value is given by:
i i i q o uv = (1)
of similar products, averaged throughout the year and over all producers70. The value given by the
uv thus represent the average value of one unit of a product from product group i. A comparison
of unit values provides the basis of the industry of origin purchasing power parities (PPPs) which will be used to compare the value of output per head in Germany and Britain. To obtain the unit value ratios (UVRs), products with similar characteristics are matched and the ratio of the unit values in both countries is taken. This unit value ratio is given by:
A i B i BA i uv uv UVR = (2)
where A is the base country and B is the country to which it is compared. The product unit value ratio indicates the relative producer price of the matched product in the two countries. UVRs need to be aggregated to derive converters for gross output. In some cases the coverage ratio in terms of total sales within an industry is rather low, which makes it difficult to assume that the UVR is representative for the whole industry. Therefore, I weighted the UVRs according to their output share in the individual industry. Then, the UVRs were weighted according to the industry share in the branch of manufacturing. Finally the UVRs were weighted according to the branch share in manufacturing as a whole.
The Gross ouput PPP (GOPPP) for industry j based on the industry of origin approach is given by:
∑
= = GO Ij i BA ij ij BA j wUVR GOPPP , 1 (3)Where i represents the matched output products in industry j; and wij is the output shares of the ith
product in industry j. In the bilateral comparison case, one can either use the weight of base country A or the weight of the other country B to obtain the GOPPP. When using the base country A, the Laspeyres gross output PPP, GOPPPjba(a) is obtained:
∑
= = GO j I i BA ij A A ij A BA j w UVR GOPPP , 1 ) ( ) ( (4)the output weight of product i in base country prices and quantities is given by wij.
When using the other countries weight as the basis of comparison the Paasche PPP is obtained. The Paasche GOPPP is given by:
∑
= = GO j I i BA ij B A ij B BA j w UVR GOPPP , 1 ) ( ) ( (5)Now, wijA( B) is the quantity weight of the other country valued at the price of the base country. In
general we can expect Laspeyres PPPs are higher than Paasche PPPs. Due to the negative relation between price and quantity, goods with high prices will have a low quantity in the own country. The quantity weights of the other country are therefore relatively large. The Laspeyres index (4) is constructed by using the weights of the base country, thus valuation of gross output at foreign quantities will tend to inflate its aggregate value. This effect is known as the “Gerschenkron effect” named after Alexander Gerschenkron who described it in detail71.
In the remainder of this thesis I will use the Fisher index, which is, the geometric average of the Paasche and Laspeyres index. The Fisher GOPPP is given by:
) ( ) ( * BAB A BA GOPPP GOPPP GOPPP = (6)
The Fisher index has some favourable properties over the other indexes. The Fisher index satisfies the country reversal test, thus changing the denominator and numerator does not alter the results72. Moreover, a Fisher price index times a Fisher quantity index gives the Fisher value index. When the price indexes are extrapolated, the Fisher index shows a smaller margin of error from the true year of extrapolation than the Paasche and Laspeyres index73.
The name PPP is slightly misleading, since the value calculated is not a real Purchasing Power Parity. It is the weighted sum of the Unit Value Ratios, which are relative producer prices. Standard PPPs are usually based on a given basket of goods, and represent the price paid in different countries for this same basket. The name UVR has been introduced by the Groningen Growth and Development Centre and will be used in the rest of this thesis.
71 Gerschenkron 1962. 72 van Ark 1990, p.30.
2.4.1 Deflation methods
Although most existing benchmark studies rely on single deflation, there exists another method to obtain benchmarks, the so-called “double deflation” method. Paige and Bombach were the first
to apply the concept of double deflation, although they called it the “double indicator method”. Fremdling, de Jong and Timmer provide a good discussion on the advantage of applying the double deflation method74. According to them the theoretically correct way to obtain a benchmark would be to use data on gross output and intermediate inputs in both countries, and convert them to a common currency using two PPP’s, one for output and one for intermediate inputs75. The method of double deflation does this by deflating gross output and intermediate inputs separately. The bias in results when single deflation is used is especially severe when there are large differences in the technical input-output coefficients of an industry between two countries.
Van Ark spelled out some methodological objections against the double deflation method76. Firstly, the Paasche and Laspeyres unit value ratios can differ substantially when the share of intermediate inputs in gross output differs between countries. Secondly, relatively small measurement errors in the price ratios of outputs or inputs tend to become magnified in the unit value ratio when intermediate inputs make up a large part of output.
Since there is a lack of data on intermediate inputs in the German and British census and there are multiple difficulties in applying the double deflation method in a consistent way, the single deflation method will be applied in this thesis.
One can also choose between deflating based on gross value and net value added. In this thesis I choose to use deflate based on gross output weights, since the German census does not provide information on net value added. Moreover, the results when using net value added can be less reliable than when using gross output figures with simple deflation. This is caused by the effect of price and input effects77.
2.5 Matching of products
As explained earlier, matching of products can be difficult in some instances, since not all products are homogenous. In some cases products have different names in the different censuses.
Another issue is that the German and British census did not use the same measurement units. The German census relies on the metric system and the British census relies on the imperial
74 Fremdling, de Jong and Timmer 2007, p.360. The double deflation method has also been applied in
studies on productivity in agriculture, see e.g. van der Meer and Yamada (1990) and Maddison and van Oostroom (1993).
75 idem, p.360. 76 van Ark 1993.
system. It was possible to convert the British data into metric units. In Table A3 in the Appendix the exact conversion factors are presented. There are however also products that cannot be matched since it is not possible to convert the measurement units to one comparable standard. For example, in certain cases the German census provides information in tons, whereas the British census provides information on the number of products. Without knowledge of the weight of the particular product, it is not possible to convert the tons to number of products or vice versa. Especially when the product actually consists of a group of different products, which might all have different weights, it becomes impossible to convert. For example, German census provides information on tons of “preserved food rubbers” in the rubber industry whereas the British census provides information on the number of rubbers to preserve food. But since these rubbers will have different sizes and hence different weights, and the exact composition of the product group is not known, I omitted these products from the comparison.
There are also cases for which no information on sales value and quantity is reported in the census. For example, in the German census there is no detailed product information when there where less than three firms active in the production of a certain good. However, in these cases the production will most probably not be very large, thus omitting the product will not cause problems. Finally, certain products are only manufactured in one of the countries and therefore cannot be matched.
There are some industries where a comparison for the post-war period makes no sense since there was no free production. The first sector were a comparison will be useless is the aircrafts sector. After the Second World War, Germany was restricted in aircraft manufacturing and only allowed to make repairs. In 1951 only 188 employees were engaged in the aircraft industry. This restriction was terminated in 1955. The shipbuilding and sea-going vessels industry was restricted until 1951, thus I did not compare this sector78.
2. 6 Labour-productivity estimates
In total, I was able to match 142 products. Table 2 below shows the coverage ratio. I was able to match 20 per cent of total manufacturing gross output of the United Kingdom, and 35 per cent of Germany’s gross output in the manufacturing sector. The coverage ratio differs across branches, which can be explained by the issues mentioned above: the heterogeneity of products, the measurement issues, and the national character of some products. The coverage ratio was highest for clothing and footwear and lowest for non-ferrous metal trades and engineering and
78 At the Potsdam Conference, the Allies decided to abolish the German armed forces as well as the
vehicles trade. The low coverage ratio in engineering is caused by the fact that Germany provides machinery in tons, and Britain in numbers. Paper trades also has a low coverage, caused by the fact that the printing products could not be matched.
Table 2: Coverage ratios in the 12 main industries
Table 3 provides an overview of the gross output PPPs for the 12 main industries when the coverage ratio of an industry is used as the weight in the GOPPP calculation. Table A4 in the Appendix provides the results for all main industries and the 26 sub-branches. The tables show the Laspeyres, Paasche and Fisher PPPs in Deutsche Mark per Pound. Obviously, this can be converted in Pounds per Deutsche Mark without influencing the results. Figure 3 shows the robustness graph. This graph shows what happens to the GOPPPs when industries with a certain coefficient of variation are included at their gross output weight79. The graph starts at 0 and runs to 100, at which point all industries are included with weight based on gross value. For example if there is an industry with a coefficient of variation of 10 percent, but a coverage ratio of 20 percent, this industry’s importance in the total manufacturing PPP gets five times as important, when we decide to include this industry based on its gross output share, instead of the coverage ratio. In
79
[
]
− − =∑
∑
= = 2 ______ 1 1 ln * 1 1 * 1 var j ij Ij i ij j Ij i ij j UVR UVR w I wUVR where the first part is simply the coverage ratio
and the second part a weighted coefficient of variation.
Coverage ratio (%) Industry United Kingdom Germany nr. of matched products Manufacturing - Total 20 35 142 Textile trades 33 89 12 Leather industry 64 35 5
Clothing and footwear trade 76 73 28
Iron and steel 23 49 11
Engineering, and Vehicle Trades 3 4 9
Non-ferrous Metals Trades 2 4 5
Food and tobacco 16 21 12
Chemicals and allied trades 19 13 22
Clay and building materials 22 29 8
Timber trades 46 16 6
Paper trades 6 9 4
figure (3) we observe a sharp decline in the Laspeyres index around 10 percent. This is caused by the influence of motor vehicles trade, which has a coefficient of variation at 9 percent and a low DM/£ PPP. Since there is no reason to believe the motor vehicle PPP is incorrect, it makes sense to include this sector at its full gross output weight. Around 20 percent we see that the gap between the three PPPs is smallest. Thus, based on Figure 3 I rerun the PPP’s while using gross output as weight when the coefficient of variation was lower than 20 percent. Table 3 below presents these results for the 12 main industries, and Table A5 in the Appendix provides the complete result.
Figure 3: Robustness check: Maximum coefficient of variation (GO weights)
2 4 6 8 10 12 14 0.0% 20.0% 40.0% 60.0% 80.0% 100.0% 120.0% 140.0%
Max imum C oeffic ient of Variation
P u rc h a s in g P o w e r P a ri ti e s ( D M /£ ) L as peyres P P P F is her P P P P aas c he P P P
Table 3: Gross output PPPs Germany and the United Kingdom (1951)
As expected, the Laspereys PPP is higher than the Paasche PPP in almost all industries except for the paper industry. Thus, the Gerschenkron effect does occur. The fact that we do not observe the effect in paper trades implies that consumer preferences are not fully reflected in price setting.
I used the Fisher PPP to convert Germany’s gross output value to Pounds. I calculated labour productivity per employee and per man-hour worked. Figure 4 below shows the results for the 12 main industries, Table A5 in the Appendix provides the result for the main industries and sub-branches. Clearly, adjusting for hours worked has a major impact on the labour productivity results. As Table 1 showed, Germany had much longer workweeks in 1951.
Figure 4 below shows that there is a substantial labour-productivity difference between Germany and the United Kingdom. In almost all industries, the United Kingdom has a higher productivity. In total manufacturing Britain’s labour-productivity is 35% higher than German labour-productivity. This substantial gap is in line with the fact that Germany’s GDP per capita was also lower than 75% of Britain’s GDP per capita. Given that the industrial sector is the driving force for economic growth in this period, this could not be true without a big productivity lead of Britain. The substantial gap also indicates that there was still scope for Germany to catch up with Britain. Without this manufacturing productivity gap, this would not be possible, since German supergrowth was industry-driven.
1. Purchasing Power Parities (DM/£)
2. Purchasing Power Parities (DM/£)
(gross output weight)
Laspeyres Paasche Fisher Laspeyres Paasche Fisher
Industry Name UK (DM/£) GE (DM/£) (DM/£) UK (DM/£) GE (DM/£) (DM/£) Manufacturing 13.59 7.61 10.17 14.52 10.21 12.17 Textile trades 6.96 5.54 6.21 6.96 5.54 6.21 Leather industry 7.47 6.00 6.70 7.47 6.00 6.70
Clothing and footwear trade 15.83 11.97 13.76 15.83 11.51 13.49
Iron and steel 11.18 6.56 8.56 11.10 6.89 8.74
Engineering, and Vehicle Trades 7.28 7.07 7.18 14.52 11.60 12.98
Non-ferrous Metals Trades 16.76 15.07 15.89 16.76 15.07 15.89
Food and tobacco 16.84 15.46 16.14 17.79 15.27 16.48
Chemicals and allied trades 20.75 16.47 18.49 20.75 16.47 18.49
Clay and building materials 17.84 9.22 12.83 16.68 10.59 13.29
Timber trades 23.55 10.65 15.84 23.55 10.65 15.84
Paper trades 11.98 12.51 12.24 11.98 12.51 12.24
Especially in the chemicals, clay and building materials, food and tobacco and miscellaneous trades the gap is large. In the next Chapter, the labour productivity benchmark will be utilized to study the proximate causes of the productivity differences.
Figure 4: Comparative labour productivity per branch in manufacturing in Germany and the United
Kingdom (1951)
0 0.5 1 1.5 2 2.5
Textile trades Leather industry clothing and footwear trade Iron and steel Engineering, and Vehicle Trades Non-ferrous Metals Trades Food and tobacco chemicals and allied trades Clay and building materials Timber trades Paper trades Miscellaneous
labour-productivity (£ per hour)
Chapter 3 Analysis
In this Chapter I analyze the labour-productivity gap between Germany and the United Kingdom. In this thesis labour-productivity is defined as gross output per working hour. This is known as “single factor productivity” or “partial productivity”80. Labour productivity can be higher in one country compared to another for a wide variety of reasons. One of the most obvious is that countries might use different amounts of other inputs, such as capital. But also labour quality can greatly differ between countries. The age, the level of education and the gender distribution of the workforce can be used as proxies for the quality. Countries might also differ in their degree of efficiency in production due to scale effects.
In causal productivity analysis based on time series, the most applied method of analysis has been growth accounting. Growth accounting is an analytical tool to measure the contribution of factor accumulation to economic growth. Usually the Cobb-Douglas production function underlying the neoclassical growth model is used81. The level of output is determined by the factors of production, which are capital and labour, and total factor productivity (TFP). The latter is the Solow residual and reflects the level of technological progress and the efficiency of factor use82. A large literature exists on growth accounting. In this thesis, I will not apply growth accounting, since the labour-productivity benchmark provides data on levels and not over time. In TFP accounting one would need to make assumptions about the labour and capital share in the economy. However, by using regression analysis the relative importance of different factors can actually be estimated. Moreover, in the postwar period, TFP in manufacturing was very large, and to decompose the residual capital and labour data is not sufficient. Regression analysis offers the option to include additional explanatory variables without the need of specifying factor shares. Since we have a benchmark at one point in time, the “technological progress” part would also need special attention, as it cannot be interpreted as an increase in technology.
Therefore, I will use regression analysis to study the productivity differential. All variables are translated into Germany to United Kingdom ratios, so that the coefficients can be interpreted as elasticities, and are directly comparable.
In the first part of this Chapter I apply shift-share analysis to determine the effect of structural differences in the two countries on labour productivity. In the second part of this Chapter, I will apply regression analysis to investigate the productivity differential between Germany and the United Kingdom.
80 Van Ark 1990, p. 5.
81 The Cobb-Douglas production function is given by: = α 1−α
3.1 Shift-Share analysis
In this Section, I will apply shift-share analysis to determine to which extent the existing productivity difference between Germany and the United Kingdom can be explained by a difference in industry structure.
Figure 5 below shows the relative sizes of the 12 main industries in terms of their share in the total hours worked in manufacturing.
Figure 5: Industry shares in total manufacturing (1951)
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Textile trades Leather industry clothing and footwear trade Iron and steel Engineering, and Vehicle Trades Non-ferrous Metals Trades Food and tobacco chemicals and allied trades Clay and building materials Timber trades Paper trades Miscellaneous
share of manufacturing employment
United Kingdom Germany
Germany had a larger share in timber trades, clay and building materials and chemicals. Especially the larger share in Timber trades and clay and building materials is interesting, since these are industries that produced goods needed for reconstruction projects. The fact that in 1951 these industries were considerably larger than in the United Kingdom confirms the possible importance of reconstruction dynamics for Germany’s supergrowth period.
The United Kingdom had a larger share of employment in engineering, motor and vehicle trade, iron and steel, textile and paper trades. Thus, an interesting question is to what extent these differences in structure cause the aggregate labour productivity gaps. Shift-share analysis can be applied to answer this question83.
Shift-share analysis has been used to distinguish between region/sector specific and inter-sectoral/inter-regional effects in accounting for aggregate growth patterns84. The first model I use is the static shift-share model85.
83
Shift-Share analysis was originally proposed by Dunn (1960) as a forecasting technique for regional growth.
(
)
(
)
(
GE)
i UK i n i GE i UK i GE i UK i n i GE i UK i GE UK LP LP LP S S S S LP LP LP − =∑
− × + +∑
− × + = = 2 1 ) ( 2 1 1 1 (7)LPi is labour productivity in pounds per hour worked in industry i, and Si is the share of industry
i’s total hours worked in the manufacturing sectors. The left-hand side of the equation presents
the overall difference in manufacturing labour productivity in pounds per hour between the United Kingdom and Germany. On the right-hand side, this productivity difference is decomposed into two effects. The first is the so-called intra-sector effect. It accounts for the differences in labour productivity within each industry, assuming equal employment shares in both countries. The second effect is the so-called shift effect. This is a structural effect, which accounts for the impact of structural advantage on aggregate productivity performance, assuming equal branch-specific labour productivity levels in both countries. I estimated the static shift-share model for the 12 main industries and for the 26 sub-branches. The result is presented in Table 4 below.
Table 4: Static shift-share analysis - Germany and the United Kingdom (1951)
Main industries Sub-branches Total gap UK-GE
(£ p.h.)
0.295 0.295
Intra-sector 98.66 % 93.04%
shift 1.33 % 6.96%.
When the 12 main industries are examined, the intra-sector effect captures almost the complete labour productivity gap and the shift effect is very small. When the 26 sub-branches are used however; the shift effect becomes larger. This is due to the fact that at the more disaggregated level the differences between the employment shares are more pronounced. Table A7 in the Appendix shows the differences in employment share at the 26 sub-branch level. At the 12 sector level, Britian has a larger share in engineering and vehicle trades than Germany. However, when we look at the more disaggregate level, it turns out that Germany has a larger share of employment in optical and precision engineering, which are both sub-branches of the engineering and vehicle industry. But since the United Kingdom had a larger share in mechanical engineering, these effects cancel out.
