Competition, Productivity and Efficiency

OECD Economic Studies No. 27. 1996/11 zyxwvutsrqponmlkjih COMPETITION. PRODUCTIVITY AND EFFICIENCY Dirk Pilat TABLE OF CONTENTS Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 The evidence on productivity gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Productivity gaps in manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Productivity gaps in services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 109 Explaining productivity levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The role of factor intensity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Other explanations for productivity differences . . . . . . . . . . . . . . . . . . . . . . . . . . . The impact of competition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 119 The determinants of productivity growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Capitat accumulation. RGD and technological diffusion . . . . . . . . . . . . . . . . . . . . . The impact of competition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 129 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131) Annex: Comparing productivity levels: measurement issues . . . . . . . . . . . . . . . . . . . 133 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 116 120 122 127 The author is grateful for helpful comments a n d suggestions from Bart van Ark. Paul Atkinson. Sveinbjorn Blondal. Martine Durand. lsrgen Elmeskov. Michael P. Feiner. Robert Ford. Peter Jarrett. Toshi Kato. Joaquim Oliveira Martins. Stefano Scarpetta and Nick Vanston . H e is i n d e b t e d to H e r v e Bource. Catherine Chapuis. Martine Levasseur. Brenda Livsey-Coates a n d Sandra Raymond for their assistance . INTRODUCTION Iro8 Productivity gains form the basis of improvements in real incomes and welfare. Slow productivity growth limits the rate at which real incomes can improve, and also increases the likelihood of conflicting demands concerning the distribution of income (Englander and Gurney, 1994). The slowdown of productivity growth in the OECD area over the past decade therefore has important ramifications. However, even though productivity growth has slowed down, productivity levels still differ substantially across the OECD area, possibly indicating an under-utilised potential for growth and catch-up with other countries. Productivity growth is influenced by a range of factors, and most studies suggest that there is no simple way t o boost it (Englander and Gurney, 1994a). Apart from some specific options, such as investment in education, RGD or infrastructure, policies to boost productivity often focus on the framework conditions for productivity growth. The degree of competition in a particular country or sector is often considered to be among the most important of such factors, since a lack of competition reduces the pressure on firms to incorporate better technology, remove organisational slack and improve productivity performance. In analysing productivity growth across countries, a distinction can be made between three different processes. First, productivity growth can result from innovative activity. For the productivity leader in a particular industry, productivity growth is to a considerable extent conditional on the development of new products and processes. Over the first two or three decades of the post-war period, most innovative activity was concentrated in the United States, as both private firms and the public sector engaged in large RGD efforts At that time, much of the RGD efforts in other countries were directed at adapting and borrowing technology developed in the United States (Englander and Gurney, 1994a). However, a s indicated below, productivity leadership in the OECD has become more diversified and innovation is currently more widespread across countries. Second, productivity growth may also be d u e t o reduced (technical) inefficiency.' An inefficient firm or industry uses more resources and factor inputs than required by a particular technology, thus tying resources to low-productivity activities and reducing the overall allocative efficiency of an economy. Exposure to a higher level of competition forces inefficient firms to restructure, freeing resources Competition, productivity and eficiency zyxwvutsr for other productive uses. This process of resource (re-)allocation, which includes the entry and exit of firms, provides an important contribution to the structural change of OECD economies (OECD, 1995). A third process that can be distinguished is technological diffusion. Firms can improve productivity by adopting production processes and products developed elsewhere (imitation). This allows them to improve productivity in a relatively straightforward way, as they do not have to engage in, often costly, innovative activity. Diffusion differs conceptually from efficiency gains, a s the latter relates to improvements made in using a given technology - even when this technology is outdated by international standards. Across countries diffusion relates to the ability of countries with low productivity levels and/or a low level of technology to incorporate {he stock of technology developed in more advanced economies (i.e. catch-up). Recent studies (Coe and Helpman, 1995; Eaton and Kortum, 1995~1,1995b; OECD, 1996) suggest that currently, even for the United States, technology developed abroad provides an important contribution t o productivity growth. Research suggests that all three processes (i.e.innovation, efficiency gains and diffusion) are influenced by competitive conditions. The diffusion of technology is promoted by openness t o international competition, and competition forces firms to incorporate new production processes and technology. Efficiency is also closely linked to competition, as weak competition may result in management and workers appropriating rents in the form of organisational slack and overstaffing. The link between innovation and competition is less clear-cut, and has been quite controversial in the literature. Currently, most studies suggest that a low degree of competition, a s expressed in high concentration rates, is not conducive to innovative activity This paper discusses the empirical evidence on cross-country productivity gaps and analyses the link between productivity and competition. It first reviews whether low productivity levels are common across the OECD area, and then tries to assess which factors, including competitive conditions, contribute t o low productivity or inefficient behaviour. Next, the paper discusses whether the rate of productivity growth is affected by weak competition. The final section draws some conclusions THE EVIDENCE ON PRODUCTIVITY GAPS Productivity gaps in manufacturing There is no obvious and simple way to measure productivity gaps, and each measure has some drawbacks. Three methods are commonly used. A first method (Van Ark and Pilat, 1993; Van Ark, forthcoming) relies on cross-country comparisons of productivity levels. An industry in a particular country may be relatively produc- 4 OECD Economic Studies No. 2 7, I996/11 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA tive by national standards, but may have a low productivity level compared with best practice abroad. The main problem for this type of international productivity comparisons is the lack of appropriate conversion factors for real output. Exchange rates are not suitable, since they are strongly influenced by monetary phenomena, and in general d o not reflect real price differences between countries. In principle, industry-specific conversion factors (or purchasing power parities - PPPs) are required that reflect these differentials across countries. Recent studies have made such conversion factors available for a large range of OECD countries (Van Ark and Wagner, 1996; Van Ark, forthcoming). Some evidence on the basis of these studies is presented in Table 1 . It reports estimates of absolute levels of labour productivity (value added per person engaged and per h o u r worked) in the manufacturing sector over the period 1960-95. The average productivity performance of the United States continues t o outrank that of the other major economies (Japan,Germany and France), although Japan in particular has made considerable productivity gains over the past decades. High labour productivity levels, in particular in terms of hours worked, are also estimated for Belgium, Finland, the Netherlands and S ~ e d e n The . ~ manufacturing sectors in these small OECD economies tend to be more specialised than those of the large countries and are, apart from Sweden, relatively capital-intensive (Pilat, 1996), contributing t o a high level of labour productivity. In the middle of the OECD productivity range are a number of foliower countries (the United Kingdom, Canada, Australia and Spain) with somewhat lower productivity levels, although in particular the United Kingdom and Spain have made substantial progress over the past decades. Canada’s manufacturing productivity level was relatively high during the 1970s and 1 9 8 0 ~but ~ its level has fallen substantially over the past decade. The bottom range of productivity performance in Table 1 is made up by Mexico and Portugal, that are still quite far behind in productivity levels Evidende presented in the table also suggests that US productivity performance improved relative to many countries in the 1 9 8 0 ~ ~ More detailed estimates of labour productivity levels, for selected manufacturing industries, are presented in Table 2 On the basis of detailed data for value added, employment and hours worked, productivity levels were estimated for 36 industries in 9 countries (Pilat, 1996) Table 2 shows benchmark estimates for 1987 and updated estimates for 1993. The countries included in the Table cover only a sample of manufacturing productivity performance within the OECD, but the relative productivity performance of these countries is relatively well documented in a range of country-specific studies LLE ! Table 2 suggests that the United States remains the productivity leader for total manufacturing, but also indicates that the leadership in particular Relative labour productivity levels in manufacturing Table 1 . zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 1960-95, United States = 100 United States lapan Germany France United Kingdom Canada 1995' I985 I973 .I 960 Value added per person engaged Value added per hour worked Value added per person engaged Value added per hour worked Value added per person engaged Value added per hour worked Value added per person engaged Value added per hour worked 100 0 25 1 60 6 47 5 48 6 69 1 100 0 19 2 56 0 45 9 45 0 68 5 100 0 55 4 72 5 66 0 52 0 81 3 100.0 48.5 76.1 70.0 53.6 82.5 100 0 78 2 75 6 72 3 54 7 82 0 100 0 68 8 86 4 85 8 59 7 84 3 100 0 74 8 63 1 70 I 59 6 68 4 100 0 72 8 81 4 85 1 69 7 69 6 49.9 70.9 58.3 32.4 88.2 54 2 83 1 63 9 34 3 85 8 23 9 48 8 68 3 56 5 06 4 71 9 31 4 07 I 50 3 81 1 82 8 51 7 04 7 00 8 Australia 50 5 52 9 50 2 Belgium2 45 6 45 3 60 7 Fin la 1zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 1d 49 2 45 9 54 4 Mexico' 24 7 26 6 34 2 The Netherlands 50 8 52 8 76 8 Portugal2 3 15 7 na 25 3 Spain2 3 29 2 I5 4 20 4 Sweden 49 8 66 0 48 5 n.a. 37.8 79.6 na 79 8 87 3 na na 73 7 26 7 40 1 75 4 96 5 na 67 6 90 3 Or latest available year The productivity estimates for these countries are directly derived from industry-of-origin studies and thus exclude PPPs from expenditure studies They are therefore not strictly comparable to the estimates for the other countries 3 Portugal/USA and Spain/USA are inferential estimates based on benchmark studies for Portugal/UK and SpainiUK that were linked to the other countries by the UK/USA comparison Based on I987 benchmark estimates from Table 2 updated with time series from Van Ark (forthcoming) and BLS (1995) Benchmark estimates for Source Finland/USA Belgium/USA and Mexico/USA are from Van Ark (forthcoming) Benchmark estimate for Portugal/UK based on Peres Lopes (1994) for Spain/USA from Van Ark [ 1995) 1 2 Table 2 . Manufacturing labour productivity levels in major OECD economies, 1987 and 1993 Value added per hour worked, leader country = 100' Industrial sectors United States lapan Germany France United Kingdom Canada Australia Netherlands Sweden Panel A: 1987 Food beverages and tobacco 32 3 100.0 Textiles, clothing and footwear 38 1 67 4 Wood products and furniture 69 5 156 Paper products and printing 97 2 47 5 Chemical products 52 9 80 8 Non-metallic mineral products 77 0 55 1 Basic metal products 94 4 100.0 Metal products 86 3 76 0 Machinery and equipment 99 0 85 6 E 1 ect r i ca 1 in a ch i n e ry 00.0 82 7 Transport equipment 96 9 100.0 Ot lizyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA e r man u fa ct u r i n g 00.0 39 4 75.3 60.1 50.2 61.2 60.1 67. I 80.3 76.3 73.8 67.6 76.7 45.3 65 3 61 7 52 4 65 0 58 0 100.0 77 0 57 3 100.0 90 0 84 9 40 1 46 1 47 4 38 I 64 7 59 5 59 9 74 2 50 6 65 4 51 3 42 1 52 5 596 546 638 81 4 680 75 1 893 70 1 642 664 69 7 583 459 422 327 532 449 56 4 57 I 42 3 61 1 358 39 3 330 95.4 100.0 100.0 62.7 100.0 97.7 80.3 68.9 59.1 93.7 47.0 47.2 57 3 60 8 64 1 100.0 72 4 75 5 93 3 100.0 66 5 75 6 55 8 67 0 Tota 1 ma i i u fact u r I ng Coefficient of variation2 00.0 66 5 78.5 80 3 59 4 76.0 51.8 98.5 82 0 I6 9 41 4 20.5 31 0 26 I 22.7 28.0 28.4 29 2 643 463 526 676 526 784 879 548 55 5 51 9 51 1 32 3 27 I 537 398 774 568 359 464 280 96.6 100.0 72 a 66 5 71 9 Panel B: 19933 Food, beverages and tobacco Textiles, clothing and footwear Wood products and furniture Paper products and printing C he in i ca I prod uct s Non-metallic mineral products Basic metal products Metal products Machinery and equipment Electrical machinery 100.0 78.3 56.0 85.0 66.9 81.8 76.8 68.9 160.0 80.3 356 41 9 176 497 526 629 78 3 676 674 890 826 703 506 566 509 739 780 672 587 540 87 0 67 I 55 3 64 3 56 9 99 4 63 3 46 4 67 3 78 9 41.7 51.5 28. I 76.4 79.7 70.6 61.4 42.5 47.9 48.2 100.0 64.5 100.0 100.0 70.4 54.0 34.6 82.2 100.0 89 4 81 0 100.0 100.0 45 2 100.0 Table 2. Manufacturing labour productivity levels in major zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA OECD economies, 1987 and I993 (cont.I Value added per hour worked, leader country = 100' industrial sectors United States lapan Germany France Unitecl Kingdom Canada Australia Netherlands Sweden Panel B: 1993? Transport equipment Other manufacturing 100.0 100.0 41 4 82 6 396 85 0 31 4 47 8 43 5 71 9 335 455 22 1 41 8 27 0 49 5 47 4 Total manufacturing 100.0 76.6 81 3 84.2 64. I 71.3 52.0 95.6 91.8 884 The procliictivity level of the leader country in each industry is indicated in bold The coefhcient of variation is the standard deviation divided by the mean expressed as a percentage It is calculated over the 35 industries for which estimates are available (see Table A 2 ) 3 Productivity levels for Germany are for 1992 Data for 1987 based on Table A2 1993 updated from 1987 benchmark using output and employment series from STAN data-base (OECD 1995a) Hours Source worked for 1993 are only available for total manufacturing (BLS 1995) Consequently the trend in hours worked at the sectoral level IS assumed to be identical to the trencl for total manufacturing i 2 OECD Economic Studies No. 2 7. 1996111 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA manufacturing industries has become more d i ~ e r s i f i e dIn . ~ 1987, the United States was the productivity leader in food products and electrical machinery, the Netherlands in textiles and chemical products, Japan in basic metal products and Sweden in metal products. By 1993, some of these relative positions had changed, with Swedish productivity performance in particular improving substantially. The productivity estimates in Table 2 cover only the 12 most important sectors. The more detailed estimates of productivity, for all 36 industries, suggest that in almost one-third of all industries, the United States remains the world productivity leader. In other industries, the leadership has passed to other countries, e.g. lapan in some heavy industries (iron and steel, shipbuilding), the Netherlands in some light and capital-intensive industries (textiles, industrial chemicals) and Canada and Sweden in some resource-related industries (non-ferrous metals and paper products, respectively). Furthermore, there appears to be a shared leadership in’ several industries, e.g. in food products (United States and the Netherlands) and in motor vehicles (United States, Japan).Table 2 also shows that the inter-sectoral variation in productivity performance, a s expressed in the coefficient of variation, is by far the largest in japan. This indicates that whereas some industries in Japan are among the world productivity leaders, others are relatively far behind (McKinsey, 1993; Van Ark and Pilat, 1993). Consequently, the average productivity level of the lapanese manufacturing sector remains below that of the United States and several other OECD economies. More evidence on productivity differences, although specific to individual countries, is available from a second approach to international comparisons, namely country-specific case studies. Two examples of such studies are those by the McKinsey Global Institute (McKinsey, 1993, 1994, 1995) and the Nati~nalInstitute of Economic and Social Research (Steedman and Wagner, 1989).Case studies have the advantage that products and firms can be carefully matched and several sources of bias at the aggregate level can be avoided. A problem with this method is that it is not always easy to generalise the results from case studies to a more aggregate level /114 Most case studies compared the productivity levels of two or more OECD countries for individual industries. In general, they found large differences in performance across the OECD area. Table 3 shows some of the evidence for selected manufacturing industries in seven OECD countries.8 In food products, the United States is the undisputed productivity leader, with particularly Japan trailing far behind. In motor vehicles, Japan and the United States are the world productivity leaders, clearly outperforming the European countries In computer equipment, there appear to be only small differences between the three major OECD countries for which data are available Comfxtition. Droductivitv and zyxwvutsrqponml effrcienw zyxwvutsrq Table 3. Productivity gaps in case studies, USA = 100 Manufacturing industries Food products.' 1990 United States lapan Germany France Spain Italy Sweden 100 0 32 0 70 0 na na na 58 0 Services Motor vehicles and equipment.2 1992 computers and parts,? 1990 100 0 1187 58 5 56 7 100 0 95 0 89 0 na na na na 40 4 39 8 79 0 B a n k ~ n g . ~ Retailing,s 1992 1990 I000 na 55 0 50 0 na 25 0 66 0 1000 44 0 89 0 87 0 73 0 na 84 0 Construction,6 1990 100 0 66 0 91 0 93 0 84 0 91 0 77 0 Value added per hour worked at industry PPPs See McKinsey (1995) Value added per employee at industry PPPs See McKinsey (1994) Productivity level for Sweden refers to passenger cars only Value added per hour worked See Baily and Gersbach (1995) 3 Transactions per employee in payments and cash withdrawal See McKinsey (1995) 4 Value added per full-time equivalent employee in general merchandise retailing See McKinsey ( 1995) Productivity 5 level for Japan refers to 1987 see McKinsey (1994) Value added per employee See McKinsey (1994) 6 Source McKinsey Global Institute (1993, 1994, 1995), Baily and Gersbach (1995) 1 2 A third approach to productivity comparisons (Caves, 1992; Mayes et al., 1994; Perelman, 1995) uses estimates of production frontiers and measures inefficiency a s the gap between observed efficiency in a particular firm and the estimated efficiency frontier of the industry to which the firm belongs (see note I ) This approach also provides some useful evidence, although it can mainly be used to analyse the existence of inefficiency within a country. Although some attempts have been made to derive estimates of international efficiency frontiers for manufacturing (Fecher and Perelman, 1992; Perelman, 1995),the data are often not particularly comparable across countries and the derived efficiency measures are of doubtful value. Five countries (Australia, Canada, Japan, the United Kingdom and the United States) have been covered by studies of domestic efficiency frontiers, and in each a significant level of inefficiency was found in many industries (Caves et al., 1992, Mayes et al., 1994) In general, this is interpreted a s each industry having a long "tail" of inefficient firms, i.e. firms that could produce substantially more output with existing inputs The results of this latter approach are more difficult to interpret than the rather straightforward comparison of output per person or per h o u r worked. For instance, an industry in a particular country may be characterised by a high level of efficiency, implying that all (or most) firms are close t o the estimated efficiency frontier for 4 OECD Economic Studies No. 2 7, I996i11 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA that industry and country. However, a high level of domestic efficiency does not exclude a low level of productivity compared to other countries, as firms in a country may be using outdated technology compared to firms in other countries. Productivity gaps in services There t h u s appears to be widespread evidence of large productivity differences in the manufacturing sector, both within countries and across countries. Given the low degree of international and domestic competition in several services, productivity might be expected to vary even more there. Data constraints limit the scope of productivity analysis for the service sector and most of the available work on international productivity comparisons therefore pertains to the manufacturing sector. However, for some services, crude comparisons of productivity across countries are possible Where productivity comparisons can be made, most of the available evidence points to a large variation in productivity in services across the OECD (Table 4 ) . In electricity, output per person differs widely between countries, with the United States, Japan, Canada and Norway having the highest productivity levels. Substantial differences in productivity performance also exist within the distribution sector, suggesting sizeable scope for productivity growth in several countries T h e highest productivity levels are estimated for the United States, France, Germany, Belgium and Luxembourg, whereas low productivity levels are estimated for Japan, the United Kingdom and some of the smaller OECD economies. In airlines, considerable differences in cost efficiency exist between countries. In this sector, the highest cost levels tend to be found in continental Europe (including Ireland), and the lowest in Australia, t h e United States, the United Kingdom, Finland and New Zealand Among the larger countries, the high costs levels of Japan and France stand out. In telecommunications, productivity is relatively low in the smaller European countries, but also in Germany. Productivity differences in postal services and railways also appear substantial, although this evidence has a somewhat historical character. Nevertheless, productivity differences appear quite large in these sectors as well. /1/6 Some evidence on productivity gaps in services can also be drawn from more detailed industry-specific comparisons across countries (Baily, 1993; McKinsey, 1992, 1994, 1995; Table 3). These studies cover the experience of selected service industries in the United States, Japan, Germany, France, Italy, the United Kingdom, Spain and Sweden. On the whole, they suggest t h e existence of considerable slack in services in many countries and considerable variation in performance across countries. Table 4. Productivity and efficiency in selected service industries Electricity United States Japan Germany France Italy United Kingdom Canada Aust ra 1 ia Austria Belgium Denmark Finland Greece Ice1a nd Ireland Luxembourg Netherlands New Zealand Norway Portugal Spain Airlines Distribution Retail Telecommunications Postal Services Railways Operating expense per available tonne kilometre, 1993 (US$) Revenue per employee, I992 (USA = 100) Mainlines Per 100 inhabitants, 1992 Average technical efficiency, 1975-88' Average technical efhciency 1986-88* 100 0 0 45 100 0 56 na 70 7 0 84 0 71 46 44 0 797 100 7 80 6 63 1 na na 0 457 0 620 0 88 68 3 72 3 0 72 89 3 52 41 0 720 0 722 59 5 77 6 68 9 45 0 850 58 4 na 0 43 0 54 0 731 0 638 0 746 73 9 59 na na 2.9 59 4 60 1 70 6 49 18 86 8 73 4 77 9 44 3.2 105 0 94 1 3.3 3.1 86 6 56 4 68 6 85 9 58 6 53 4 48 0 43 58 54 2.5 37 1 62 2 36 9 44 n.a. n.a. 1i.a. 3. I 38 3 68 7 75 1 60 3 39 6 52 7 54 31 101 3 130 1 132 7 61 54 8 88 0 3.4 95 2 77 8 85 8 0 35 I 08 1 04 1 00 0 44 0 47 na I 46 na 0 48 0 44 65 2 49 44 8.0 42 3 92 9 1 10 52 2 53 0 893 na 0 600 0 732 0 198 0 387 na 0 355 0 787 0 924 na 0 630 0 630 0 523 0 653 0 564 na 0 731 0 562 0 797 na 0 516 1.2 3.3 45 4 52 8 0 83 58 9 31 77 6 45 7 0 66 74 2 40 na na 0 692 0 647 Gigawatt-hour per person engaged, I993 Distribution GDP per person engaged, 1990 (USA = 100) 8.2 100 0 6.3 2.2 60 3 78 5 3.8 96 6 95 3 94 8 2.2 5.5 1.6 per employee, I990 fUSA = loo) na 0 594 Tau,? zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 4 . Product.. ty and efficiency in selected service industries (cont.) ~~ Electricity Gigawatt- hou r per person engaged, I993 Ai rl i nes Distrr bution Distribution GDP per person engaged I990 (USA = 100) per employee I990 (USA = 100) Operating expense per available tonne kilometre 1993 (US$) Telecommunications Revenue per employee 1992 Mainlines per 100 inhabitants (USA = 100) I992 Railways Average technical efficiency 1975-88' 1 I Sweden Switzerland n.a. Turkey n.a. 1. 5.6 50.4 100. I 27.4 Average techn ica I efhciency I 986-882 68 0.755 0 662 61 0.574 0.736 I6 n.a. 0.769 Defined as output relative to inputs, where output is the sum of the number of letters delivered and the financial operations performed, inputs include employees, number of motor vehicles and number of postal offices used lsee Perelman and Pestieau (1994) for details]. 2. See note 1 . Output is the combination of gross hauled tonne-kilometres by freight trains and gross hauled tonne-kilometres by passenger trains The inputs are engines and railcars, employment, and electrified and non-electrified lines lsee Pestieau ( 1993)l. Source: Electricity based on OECDAEA (1995) and national sources for employment; Distribution GDP per person based on OECD National Accounts and national sources, converted with 1990 PPP for expenditure on goods from OECD (1993); Retail sales per employee based on EC (1993) and national sources, converted with same PPP; Airlines based on data provided by the Institute of Air Transport, Paris, for major airline companies; Telecommunications from OECD (1995b); Postal services from Perelman and Pestieau (1994); Railways from Pestieau (1993). Competition, productivity and efficiency zyxwvutsrq EXPLAl NzyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA I N G PRODUCTIVITY LEVELS The role of factor intensity Part of the difference in manufacturing productivity between countries can be explained by differences in factor use, reflecting differences in factor endowments and relative factor prices (Salter, 1966). Firms in countries such as Mexico and Portugal are faced with relatively low labour costs and consequently choose relatively labour-intensive production techniques, leading to low levels of labour productivity (see Table 1 ).zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Best practice technologies from more advanced countries may be of little relevance to these firms, a s such technologies are often based on a different set of factor prices. Table 5 Explanations of productivity gaps in manufacturing, 1989 All levels relative to the United States Value added per hour worked Relative capital intensity Value added per unit of fixed capital Total factor productivity (TFPI' Relative level of workforce q ua 1 i fi ca t ions TFP adlusted for labour force skills' TFP adlusted for skills and industrial structure? Japan Germany France United Kingdom United States 73 9 84 0 102 2 82 3 83 6 91 I 132 2 69 0 84 1 62 0 75 6 82 0 66 8 100 0 100 0 100 0 100 0 97 7 83 o 98 5 84 5 95 9 86 5 95 2 69 1 100 0 100 0 85 7 78 9 84 8 69 2 100 0 76 5 96 6 81 I Value added per hour worked adjusted for capital per worker 1 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 2 TFP adlusted for educational qualihcations of the manufacturing workforce For lapan the adlustment o n l y takes account of general qualihcations for the other countries the adlustment also incorporates vocational qualifications Workforce qualihcations are for 1987 TFP adlusted for the composition of manufacturing industrial composition based on 1987 data 3 Source Labour productivity levels are from Van Ark (1996) Adjustment factors are based on Van Ark and Pilat (1993) for japan and Germany relative to the United States and on Van Ark (1993) for France and the United Kingdom relative to the United States To some extent, differences in factor prices also affect productivity differentials between countries with more similar factor endowments, such as those reported in Table 2. Thus, part of the difference in labour productivity across countries can be explained by differences in capital intensity and capital productivity For instance, japan and the United Kingdom have relatively low levels of capital intensity in their manufacturing sector and relatively high levels of capital productivity (Table 5 ) . For these countries, capital intensity explains a substantial part of the labour productiv- 1/9/ OECD Economic Studies No. 2 7, I996I11 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA ity gap with the United States, more than 25 per cent of the gap between Japan and the United States, and almost 13 per cent of the gap between the United Kingdom and the United States However, for France, Canada and the Netherlands, which have more capital-intensive manufacturing industries than the United States (Pilat, 1996), an adjustment for capital intensity does not help to explain the labour productivity gap with the United States. In general, this implies that capital productivity in the manufacturing sector of these countries is relatively low. However, the comparison of real capital stocks and capital productivity across countries is more difficult than the comparison of real output, implying that these numbers should be evaluated with care.’ An adjustment for the average educational skills of the manufacturing work force explains only little of the productivity gap. Although the average IeveI of schooling in the United States is among the highest in the OECD (Englander and Gurney, 1994a),the average skill level of its manufacturing work force - measured by the qualification levels of manufacturing workers - is not very different from that of other major OECD countries (Van Ark and Pilat, 1993; Table 5). In addition, the experience with transplant production suggests that companies are able to match the productivity of their parent company abroad by using local labour, suggesting that educational differences are not a binding constraint for the achievement of high productivity, as appropriate in-company training can reduce such differences (McKhsey, 1993; Baily and Gersbach, 1995). Other explanations for productivity differences Differences in labour productivity may also be the result of structural differences, i.e. differences in the composition of output within a particutar industry or sector. The McKinsey studies (McKinsey, 1993) suggests that this factor plays only a limited role in most industries. However, for some of the industries shown in Annex Table 2 this factor probably plays an important role For instance, Japan is one of the few countries in the OECD that produces supertankers, contributing to a high level of labour productivity in shipbuilding. In addition, the productivity leaders in the aircraft industry, the United States and France, are the main producers of large passenger aircraft. At the aggregate level, an adjustment for industrial structure does not contribute much t o the overall explanation of productivity gaps, however (Table 5). In fact, an adjustment for industrial structure increases Germany’s productivity gap with the United States, a s the industrial structure of Germany is more geared towards industries with high absolute levels of labour productivity (Van Ark and Pilat, 1993). b?@. The combined adjustments for capital intensity, labour force qualifcatjons and industrial structure explain more than 45 per cent of the Japan-US productivity gap, and almost 20 per cent of the UK-US productivity gap. For Germany and France, the * Competition, productivity and eficiency zyxwvutsrqp combined adjustments fail to provide any explanation for the productivity gap with the United States, although the adjustment for educational skills reduces the productivity gap somewhat. More evidence on explanatory factors for productivity differences, although somewhat specific to individual countries, is available from the country-specific studies mentioned above, and primarily from the McKinsey studies (McKinsey, 1993, 1994, 1995). These studies generally confirm that differences in capital and skill intensity do not explain much of the productivity gaps in manufacturing. They also suggest that access to technology is not a major explanatory factor for productivity differentials across OECD countries. Much technology is embodied in capital goods, which tend to be easily accessible in the world market. There are, however, major differences in the degree to which the latest technology is incorporated in the production process, suggesting that new technology is only slowly diffused across countries. Economies of scale do appear to play some role, at least for some countries For japan relative to the Unjted States, Van Ark and Piiat 1993) found that the small size of establishments in many lapanese industries contributed substantially to the low level of average productivity. Similar evidence was presented by the McKinsey work (Baily and Gersbach, 1995). These studies found that sub-optimal scale and craft production processes still made up a substantial part of Japanese (e.g.in food manufacturing) and German industry (e.g.in beer production and in metalworking), contributing to low productivity levels in these industries. Firms are increasingly looking across national borders to analyse the performance of their major competitors and of the productivity leaders in a particular industry. This is particularly the case in global markets, where firms are faced with a high degree of competition. This ”benchmarking” allows them to look at best practice and sets standards for their own performance. In this context, the McKinsey studies found that much of the differences across countries with regards to productivity performance actually resulted from the “organisation of functions and tasks”.I2These differences are often the result of an accumulation of small improvements over a long period of time, regarding both workfloor organisation and the management of the firm (Baily and Gersbach, 1995). If differences in productivity levels across countries are not just the result of differences in factor endowments or structural effects, then the gap in productivity levels between countries can partly be taken a s the gap between best available (average) practice and average implemented practice in a particular country This would suggest the existence of considerable scope for catch-up In addition, the large variation in productivity performance in some countries indicates that the catch-up process with US productivity levels has not been uniform across industries, suggesting that productivity growth in some sectors may have suffered from structural rigidities, other than the availability of technoiogy (Englander and 4 OECD zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Economic Studies No. 27. /996/11 Gurney, 199423). T h e evidence on productivity differentials also indicates that even within the United States, catch-up possibilities may exist in some sectors. Where services are concerned, country-specific factors appear to affect productivity levels to some degree For instance, in electricity, favourable resource endowments in some countries (e.g. Canada and Norway) allow a high share of hydropower in electricity production, contributing to high productivity levels in these countries. In distribution, a substantial part of the differences in productivity appears related to structural characteristics (population density, land prices, etc.). In airlines, specific factors such as higher fuel costs, shorter stage lengths and higher fly-over costs reduce the efficiency of European airlines relative to US-based carriers ( H s j , Kato and Pilat, 1996) The impact of competition The impact of competition on productivity is not so easy to evaluate The degree of competition in a particular industry is difficult to measure and is determined by many different factors. Competition also does not affect productivity in a direct and easily measurable way, a s tends to be the case with production factors such as capital or technology. Rather, it is an important determinant of the conditions under which productivity growth occurs and under which high productivity levels may emerge A first look at the link between productivity levels and competition is provided in Table 6. It shows pooled correlations between a set of competition-related variables and labour productivity levels, for 9 countries and 36 industries The correlations distinguish between different types of industries, based o n a typology by market structure (Oliveira Martins, Scarpetta and Pilat, 1997) This typology allows a distinction between different types of competitive behaviour. For instance, fragmented, homogeneous industries are most likely to be characterised by perfect competition. Firms in these industries are typically small and the goods produced by these industries are relatively homogeneous Examples of such industries include food products and textiles. On the other extreme, segmented, differentiated industries are mostly made up of large firms, producing highly sophisticated and differentiated goods. Examples of such industries include drugs and medicines, as well as electrical and computer equipment. First, as expected, relative levels of capital intensity are positively and significantly related to labour productivity levels Concentration rates are not correlated to productivity levels in fragmented industries, but there appears to be a significant negative link for segmented industries, suggesting that high concentration is not conducive to productivity in these industries Entry rates have a positive impact on productivity levels, particularly in fragmented, homogeneous industries, where entry /122zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA is easiest and firms are relatively small Table 6 Correlations between productivity levels and structural variables' T-statistics in parenthesis Trade Variables Market structure groups 2 Capital intensity level Concentration rate Entry rate Export intensity Import penetration MFN tariff rates Rate of core NTBs Fragmented, homogeneous 0.05 (0.36) 0.44 (3.97) Fragmented, differentiated 0. I6 (0.78) ( I .70) -0 20 (-1 2 1 ) -0 30 (-1 86) -0 10 (-0 58) 0 41 12 68) 0.3 1 -0 09 -0 19 0 06 0 20 ( 2 05) zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCB (-0 85) (-I 95) (0 57) Segmented, homogeneous -0.35 (-2.51 ) 0.15 (1.19) 0 24 ( 2 17) 0 07 (0 59) -0 18 (-1 57) -0 05 (-0 40) Segmented, differentiated -0.31 (-2.12) 0.36 (2.90) 0 04 ( 0 40) -0 15 (-1 44) 0 03 (0 32) 0 23 ( 2 18) I. The Table shows correlations between structural variables and the pooled sample of labour productivity estimates For all 9 countries and 36 industries For which estimates could be derived. Concentration rates (see Van Ark and Monnikhof. 1996) are only available for the United States, lapan, Germany, France and the United Kingdom Entry rates are only available for the United States, lapan, Germany, the United Kingdom, Canada and the Netherlands. 2. Detail on the classification of industries by market structure is available in Oliveira Martins, Scarpetta and Pilat (1997). Source. Calculations on the basis of labour productivity levels reported in Annex Table A2; capital intensity levels from Pilat (1996);concentration rates from Van Ark and Monnikhof (1996); entry rates from Schwalbach (19911, Management and Coordination Agency (1989/90 and 1993/94) and Kleijweg and Lever (1994), export intensity and import penetration from STAN database (OECD, 1995a): tariff barriers and NTBs from OECD (19960). OECD Economic zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Studies No. 2 7. I996/11 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA The trade variables also provide some interesting results. Export intensity and tariff barriers are only correlated with productivity levels in homogeneous industries, i.e, those industries where price competition is most important. Though not significant, the negative link between import intensity and productivity levels is more difficult t o interpret, although it could indicate that import penetration increases if industries in a particular country can not match high productivity levels abroad. Non-tariff barriers (NTBs) appear positively linked to high productivity levels, but only in differentiated industries. This result is surprising and difficult to interpret, as NTBs are primarily used in declining industries, most of which tend to produce homogeneous goods. The correlation analysis is somewhat elaborated in the regression analysis of Table 7. It appears that some of the variation in manufacturing labour productivity levels seems related to the exposure of sectors to international competition, a s measured by tariff barriers and export intensity, while, a s expected, a much larger part appears due to differences in capital intensity T h e regressions only explain about half of the variation in productivity levels, however. This is not surprising, a s a complete model of productivity levels is difficult to test empirically. Many relevant variables, for instance work force qualifications or the degree of foreign direct investment by industry, are not available at a sufficient level of detail and are thus difficult to integrate in a regression analysis In the McKinsey work, the degree t o which firms implement modern technology was directly related to their exposure to competition A firm in a sheltered market has few incentives to choose an efficient technology and reduce resource use, but can spend the rents it earns on technical inefficiency and organisational slack. For Germany, lapan and the United States, Baily and Gersbach (1995) argued that differences in the use of modern technology were strongly correlated with the extent to which entire industries were exposed to competition with best-practice firms, either by international trade or by competition with transplant companies. In general, domestic competition by itself was insufficient to bring firms - and hence entire industries - up to global best practice productivity levels. The work on efficiency frontier measurement also provides some links between competition variables and inefficient behaviour (Caves, et al., 1992). Of particular interest are two types of explanations: - Competitive conditions. In most countries efficiency within an industry declines beyond a certain level of concentration, suggesting that high levels of concentration are detrimental to efficiency. In Japan and Canada, exportoriented industries were found to be more efficient than import-competing industries, whereas import competition favourably affected efficiency in the United States and the United Kingdom Furthermore, tariff protection (in Japan and in Australia) and entry-restricting regulations (in japan) were found t o have a negative impact on efficiency. T h i s suggests that Table 7. Regressions explaining labour productivity levels in manufacturing, I987 Dependent variable is real value added per hour worked (in logs), t-statistics in parenthesis' Capital Constant intens it y level (in logs) 0 477 30 1 0 267 0 504 300 0 260 0 510 297 0 258 -0 250 (-3 24) 0 522 300 0 255 -0 238 (-3 09) 0 526 297 0 254 Import penetration R2 (in logs) (in logs) ( i n logs) (adi) 3 414 17 84) 0 189 (4 80) Equation 2 3 773 1831) 0 154 ( 3 92) 0 091 (4 06) Equation 3 3 866 18 53) 0 145 ( 3 69) 0 086 (384) 3 829 ( I 8 85) 0 162 (4 20) 0 103 ( 4 63) 3 916 (19 02) 0 154 ( 3 98) 0 098 ( 4 39) Equation 5 Standard errors of regression Tariff measures Equation I Equation 4 Nob Export intensity -0 065 (-2 14) -0 056 (-1 88) 1 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA The eqiiations include hxeci country effectsand sectoral effects They are based on the pooled sample of productivity estimates for 9 countries and 36 industries (see Annex) See source note to Table 6 Source OECD Economic Studies No. 2 7, I996111 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA competition reduces the spread in performance within an industry, probably by removing the most inefficient firms and by boosting performance in other firms. In general, domestic competition appears to play a larger role than international competition in promoting efficiency (i.e. eliminating intrasector efficiency differences) within a given sector in an individual country. - Organisational and managerial influences. Although difficult t o measure, t h e ultimate source of inefficiency is often related to the management of a particular firm. Only some aspects of this factor have been analysed, however. For instance, Caves and Burton (1992) found that many U S firms had diversified their activities too much in the 197Os, leading to a decline in efficiency. Apart from these competition-related reasons, other explanations for inefficiency in these studies of efficiency frontiers were provided by industrial dynamics (fast-growing industries are likely to show a larger variation in perFormance), spatial disparities (efficiency is likely to difFer in a geographically diversified market) and heterogeneity of products (which implies that the calculated efficiency frontier may not be appropriate for the industry as a whole) For services, the link between competition and productivity can often only be made in a rather descriptive way, as few hard indicators of competition are available Within private services, the profit motive provides incentives for managers to enhance productivity, although regulations can limit the impact of competition. Regulations may also prevent firms from using business practices developed in other countries (OECD, 1 9 9 5 ~ )For . sectors controlled by public enterprises, profitmaximising behaviour is likely to play a more limited role in managers’ decisionmaking. Across different sectors, there are some indications that competition matters. In electricity, the variation in productivity tends t o be large even for countries with a similar composition of supply, suggesting that pure efficiency differences may play some role. For instance, productivity levels in the United States are substantially higher than in many other countries, even though the bulk of U S electricity production is based on combustible fuels In distribution, part of the variation in productivity seems due to a high degree of regulation in some countries, principally limitations on large-scale establishments and restrictive zoning laws ( H s j et al., 1996) In airlines, the evidence suggests that deregulated markets (primarily the United States, the United Kingdom, Canada, Australia and New Zealand) have much lower costs and are much more efficient than regulated markets (mainly those of many countries in continental Europe) Within public enterprises and in particular in sectors that were, until recently, j126 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA generally considered to be natural monopolies (e.g telecommunications, railways, Competition, productivity and eficiency zyxwvutsrqp postal services), the link between efficiency and competition is probably stronger The lack of competition in these sectors and the high degree of public ownership have reduced incentives for cost-minimisation and efficiency improvements and have often led to a substantial degree of overstaffing (Pera, 1989). Furthermore, public enterprises tend to have lower internal efficiency than private enterprises (OECD, 1994). In telecommunications, technological improvements and the rationalisation of activities, partly due to increased out-sourcing, have substantially increased productivity. This trend, and the ensuing fall in costs and prices have been strongest in countries with competitive telecommunications markets (OECD, 1995b) For European railways, the evidence (Pestieau, 1993) suggests that some of the efficiency differences between companies (Table 4 ) are related to t h e degree of autonomy under which railways operate. Research on postal services within Europe (Perelman and Pestieau, 1994; Table 4 ) also suggests a link between efficiency and the degree of regulation and managerial autonomy of postal service companies Case studies of service productivity also often relate the existence of slack to a lack of competition (Baily, 1993) For instance, case studies of t h e banking industry suggested that this sector was characterised by considerable inefficiency, and few incentives to adopt new technology and available innovations before competitive pressures increased. Similar conclusions were drawn from case studies on airlines, construct ion and telecom m U n i ca t ions. THE DETERMlNANTS OF PRODUCTIVITY GROWTH Capital accumulation, R I D and technological diffusion The analysis of productivity levels generates useful insights in cross-country productivity differences and the links with competition The analysis of productivity growth is a more standard tool of economic analysis and may also provide useful views on the impact of competition Most research suggests that capital accurnulation, RGD expenditure and human capital accumulation are the prime drivers of productivity growth (Englander and Gurney, 1994a; OECD, 1996). Unfortunately, human capital is difficult t o integrate in detailed sectoral productivity analysis, due to data constraints. However, a regression analysis of manufacturing productivity growth at the sectoral level suggests that labour productivity growth is indeed positively affected by the growth of capital intensity, and also by the growth of the available RCD stock (RGD expenditure, accumulated over 5 years) per person engaged (Table 8) In recent years, much attention has focused on the role of technological diffusion Recent studies (Coe and Helpman, 1 9 9 5 ; OECD, 1995; Eaton and Kortum, 1995a, 1995b) suggest an important role of foreign RGD in productivity growth, rzzyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA lk zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Table 8. Estimates of labour productivity growth for manufacturing industries, 1981-90 Dependent variable is real growth of value added per person employed, t-statistics in parenthesis ~ ~~ Competition variables Growth of capital i n te ns i ty Growth of RGD stock per person Tariff measures ( i n logs) Equation l 2 0.370 (5.69) 0 196 ( 7 37) Equation 22 0.330 (4.93) 0 200 ( 7 48) -0 262 (-1 02) Equation 33 0.36 1 (4.89) 0 287 ( 7 64) -0 761 (-3 08) Equation 42 0.48 1 (6.17) Equation 52 0.3 15 (4.01) Nob' lad') Standard error of equation 0 664 ( 3 29) 0 536 2 23 315 0 623 ( 3 02) 0 529 2 22 31 1 0 491 2 32 I92 - 1 363 (-3 29) 0 513 2 29 25 I -0 759 (-I 92) 0 566 2 13 242 Export intensity ( i n logs) Entry rate ( i n logs) ( i n logs) I058 ( 2 13) 0 889 (361) 0 256 ( 7 00) 1974 Labour productivity level R2 There are some differences in the coverage of the equations, due to availability of the basic data. Data on RGD stocks are not available for Australia, Belgiuin and Norway, while entry rates are only available for the United States, lapan, Germany, the United Kingdom, Canada, Belgium, the Netherlands and Norway, 1974 procluctivity levels could not be derived for Australia 2 Equations include fixecl country and fixecl sectoral effects Equation includes only fixed country effects Fixed sectoral effects were not significant 3 Source Calculations are based on STAN data-base, RGD stocks are calculated on the basis of RCD flows from OECD's ANBERD data-base (OECD. 199513).other variables are derived from sources cliioted i n Table 6 1 Competition, productivity and efficiency zyxwvutsrqp attesting t o the importance of diffusion. A recent OECD study (OECD, 1996) concluded that: technology diffusion often accounted for more than half of total factor productivity growth; that its contribution was typically larger than that of direct RGD; and that the role of technological diffusion increased from the 1970s to the 1980s. Not surprisingly, this study also found that technological diffusion was particularly important for smaller countries. Productivity growth also depends on the starting point, with low initial levels permitting faster growth This catch-up factor has given rise to a substantial literature, which tends to conclude that countries can catch-up with the productivity level of the leader country, providing that they have a sufficient stock of education and basic knowledge to absorb technology from abroad (Abramovitz, 1989). Among OECD countries, catch-up and convergence of income and productivity levels has been observed at the macro-economic level. Table 8 (equations 4 and 5) confirms that some element of catch-up is also at work at the industry level in the manufacturing sector, with low initial productivity contributing to more rapid productivity growth. The impact of competition As indicated before, the impact of competition o n productivity growth is likely to be more indirect. By allowing inefficiencies to persist, weak competition may affect productivity growth. A lack of competition may also put insufficient pressure on management to improve productivity performance and incorporate new technology, and thus contribute to a productivity gap with best practice The regression analysis reported in Table 8 appears t o confirm that the degree of competition has some impact o n productivity growth. Tariff measures appear to have a negative effect on labour productivity growth, whereas export intensity affects productivity growth positively. These results support the view that exposure to international competition promotes productivity growth. There are a number of alternative interpretations of the link between exports and productivity per-formance, however (Englander and Gurney, 1994a). For instance, competition on the international market can contribute to cost minimisation, but exports may also allow specialisation and economies of scale It also appears that a dynamic product market, a s measured by entry rates, provides a positive contribution to productivity growth. High entry (and exit) rates could ensure that only the best (and most productive) firms survive the competitive process, thus promoting productivity growth in an industry (Nickell, 1996) The impact of competition on productivity growth is confirmed by a number OF other studies, many based on industry- or firm-level panel data for individual countries (e.g. Haskel, 1991). These studies found that high degrees of market concentration and market share have an adverse effect on the level of total factor 4 OECD Economic Studies No. 27, I996/11 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA productivity. A more recent study for the United Kingdom (Nickell, 1996) confirmed this result, but also found that competition, measured by an increase in the number of competitors or lower levels of rents, is associated with higher total factor productivity growth. A reduction in technical inefficiency within countries may also contribute to TFP growth. Although few studies have covered this aspect, the available evidence (Perelman, 1995), suggests that technical efficiency in the manufacturing sector of most OECD countries has actually declined over the period 1970-87 (i.e.the spread in productivity performance within industries has increased). According to this study, the main exceptions to this were Belgium and Japan,where improvements in technical efficiency provided a sizeable contribution t o TFP growth over this period. For services, the available evidence suggests a positive contribution of efficiency improvements t o TFP growth in most OECD countries over the period 1971-86 (Fecher-and Perelman, 1992). The determinants of productivity growth may also differ according to market structure type (Table 9). It appears that the growth of capital intensity contributes to productivity growth in segmented, homogeneous industries, i.e. industries with high s u n k costs, but has no significant ef€ect in the other sectors. The growth of RGD stocks contributes to productivity growth throughout all sectors, but its effect is by far the strongest in segmented, differentiated industries, although it is also substantial in fragmented, differentiated industries. The negative contribution of tariff measures to productivity growth is strongest in homogeneous industries, where price competition is likely to be most intense High entry rates provide a positive, and significant contribution to productivity growth, except in segmented, differentiated industries which are most likely t o be characterised by oiigopolistic ma rket st ru ct u res . The link between competition and productivity growth is perhaps most clearly demonstrated by the experiences with service sector deregulation in many OECD countries (Winston, 1993; H 0 j , Kato and Pilat, 1996). For instance, the deregulation of the US airline market since 1978 and that of the United Kingdom over the 1980s led to a sharp restructuring of the industry and a large increase in productivity. The deregulation of road freight transport in many OECD countries {OECD, 1990) and of the telecommunications industry in the United States, the United Kingdom and Japan (Harris, et al., 1995) led t o similar experiences CONCLUDING REMARKS E Although the evidence is scattered and incomplete, a number of conclusions emerge from the discussion above. First, it appears that inefficiency and low productivity levels are widespread in both manufacturing and services, and throughout the OECD area, suggesting a substantial potential for further productivity growth in Table 9 Estimates of labour productivity growth for manufacturing, by market structure, 198 1-90' Dependent variable is real growth of value added per person employed, t-statistics in parenthesis competition variables Constant Growth of capital intensity Growth of RGD stock per person Tariff measures (in logs) Entry rates (in logs) R2 (adi) Nob A11 industries 2 0.409 (5.72) 0.216 (5.67) 1.862 (5.25) 0 386 i92 Fragmented, homogeneous 0.136 (1.28) 0.037 (0.75) 0.976 (2.26) 0.096 63 Fragmented, differentiated 0.207 (0.62) 0.352 (2.94) 3.635 (2.60) 0 398 22 Segmented, homogeneous 0.502 (4.12) 0. I98 (2.34) 1.91 1 (2.85) 0.353 66 Segmented, differentiated 0.210 ( I .36) 0.399 (5.65) 0.633 (0.60) 0.573 41 Equations are based on pooled data for the United States, Japan. Germany, the United Kingdom, Canada and the Netherlands. This equation is similar to equation 3 reported in Table 8, but excludes fixed country effects. Source. Calculations based on STAN database (OECD. 1995a). 1. 2. OECD Economic Studies No. 2 7. I996I11 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA many countries. The great variation in the speed of catch-up across industries may indicate that structural factors inhibit productivity growth in some sectors. Second, the variation in productivity levels and growth rates across countries appears to some extent related to the degree of competition facing industries and sectors in different countries. International competition, both in the form of trade and of direct foreign investment, appears to be an important element in achieving high levels of efficiency, while case studies suggest that the highest levels of efficiency are achieved by industries competing with best (global) practice. Productivity growth in manufacturing appears positively affected by open borders, a high export intensity and favourable entry conditions Openness also allows firms to learn from and benchmark their performance against that of their international com pet itors There are also some signs that high entry rates are conducive to productivity and that a high degree of concentration is not Furthermore, sectors with a rapid growth of RGD stocks enjoy more rapid productivity growth. Industry-specific catchup appears to play a role in explaining productivity growth, indicating the importance of technological diffusion and openness. In service sectors, governmentimposed regulations are often an important restriction on competition, preventing entry and reducing the benefits of competition. Regulatory reform in many service sectors has led to an increase in competition and almost invariably to higher productivity growth. Finally, the evidence from studies on efficiency frontiers suggests that a low degree of competition within a country, as indicated by low entry rates or a high degree of concentration, is likely to lead to a high variation in efficiency and productive performance and consequently to sub-optimal average productivity. However, if inefficient behaviour results from a lack of competition within a market or a high degree of regulation, the policy implications are more obvious than if inefficiencies result from other factors such as differences in the product mix or a geographically diversified market Annex COMPARING PRODUCTIVITY LEVELS: MEASUREMENT ISSUES This annex briefly describes the methodology and data that were used to compute the labour productivity levels for manufacturing presented in the main text of the paper.I3 Productivity estimates were calculated for 9 countries, with industrydetail for 36 sectors. Two main problems had to be solved. The first relates to the conversion of value added in national currencies to a common currency. The second problem concerns the basic sources on value added, employment and hours worked. In principle, the appropriate conversion factors for productivity comparisons need to be derived from a comparison of producer prices for specific goods. Such prices are sampled for most countries for the construction of the overall producer price index, but these data are often not available for outside analysis and may be difficult t o compare across countries Another source of producer price information -at least for manufactured products - is the census of manufacturing industries. Most countries publish such a census, which shows production values and output quantities for a range of products, in principle allowing the comparison of producer prices and the derivation of appropriate conversion factors. This approach, the “industry-of-origin” approach, has been used in a range of studies, starting with some early work at OEEC (Paige and Bombach, 1959). Recently, most efforts in this area have been made by a group of researchers at the University of Groningen (Van Ark and Pilat, 1993; Van Ark and Wagner, 1996; Van Ark, forthcoming). The results of this approach have been scrutinised in a number of studies The results of a comparison for Germany, Japan and the United States (Van Ark and Pilat, 1993) were carefully checked by work at McKinsey (McKinsey, 1993; Baily and Gersbach, 1995; Gersbach and Van Ark, 1995). The McKinsey work profited from detailed knowledge by industry experts and price information at the firm level. Substantial changes were made to some price comparisons (mainly for investment goods), but price comparisons for more homogeneous products (iron and steel, ,131 OECD Economic Studies No. 27, I996111 beer) were hardly affected. The overall perspective on Japanese and German productivity performance changed little, however. There are a number of problems involved in using the industry-of-origin approach: - The “prices” derived from the manufacturing census relate to average prices or “unit values” (i.e.values divided by quantities). If a country is producing a wide range of qualities and varieties of a particular good, the “price” is rather crude for comparative purposes. In a cross-country context, quality differences between countries may not be properly accounted for. This issue is less likely t o be a problem for industries producing relatively homogeneous goods - Unit values are available o n l y for a sample of goods, and can be compared among countries for an even smaller sample, partly because of confidentiality problems. In addition, the production structure of countries tends to be far less comparable than the expenditure structure. Both problems imply that the unit values only cover part of the manufacturing sector, and that an aggregation procedure is required to cover manufacturing a s a whole. - The third major problem is double deflation Comparisons of labour productivity or total factor productivity are generally based o n value added by industry, which implies, in principle, that conversion measures for both output and intermediate input are required In practice, conversion factors for intermediate input are very difficult to derive in a cross-country context Most studies have therefore tended to apply the conversion factors at the producer level directly to value added (i.e. single deflation, see Van Ark, forthcoming) 1134 Although the production approach is theoretically the correct approach to sectoral productivity comparisons, it is therefore not without some measurement problems. Some authors have therefore used the more widely available price information (purchasing power parities or PPPs) on the expenditure side (Jorgenson and Kuroda, 1992; Kuroda, 1996).This type of information is available for almost all OECD Member countries at a fairly dis-aggregated level and new comparisons of this type are made on a regular basis. Extensive data sets are available for 1985 (OECD, 1987), 1990 (OECD, 1993) and 1993 (OECD, 1996a). T h e 1990 price comparisons covered about 2500 goods and services, and detailed comparisons are available for about 220 “basic headings”. The price comparisons are based o n detailed product descriptions, which generally ensures a rather high quality of the price comparisons. Competition, Productivity and Efficiency Where productivity measurement is concerned, these price comparisons are rather problematic, however There are five problems in using expenditure PPPs for sectoral productivity comparisons: - Distribution and transport margins. PPPs on the expenditure side are based on comparisons of prices at the retail (for consumer goods) or wholesale (for investment goods) level This implies that distribution and transport margins are added to the producer price, and that cross-country differences in the size of these margins affect the estimated price level - Indirect taxes less subsidies. Prices at the expenditure level include indirect taxes less subsidies, implying that differences in VAT and other indirect taxes (duties) across countries affect the measurement of the relative price level - International trade. International productivity comparisons should be based on the output produced in a country. However, part of this output is exported and not counted in comparisons of expenditure prices, while imported goods are taken into account in expenditure comparisons, but should be excluded for producer price comparisons - Intermediate goods. Expenditure comparisons only cover goods entering final expenditure (see above). Intermediate goods, that form the bulk of output in many sectors of the economy, are not covered. - Double deflation. Even after "peeling off" distribution margins and net indirect taxes, and after adjusting for international trade, the prices derived still refer t o output only. No information is available on prices of intermediate goods that would allow double deflation Providing data are available on the distribution margins by country (and preferably by sector), it is fairly simple to adjust for distribution margins (Hooper and Vrankovich, 1995). It is also fairly simple to adjust for taxes and subsidies Adjusting for international trade is not so easy (Hooper and Vrankovich, 1995), a s information on the price levels of exports, imports and domestic production is required. Although some simplifying assumptions can be made, no simple solution is available and many expenditure studies (lorgenson and Kuroda, 1992) have not addressed this problem. The fourth problem, that of intermediate goods, can not be addressed if only expenditure PPPs are used Most studies of this type use price ratios of other goods to f i l l these gaps. The fifth problem, that of double deflation, is inherent to international comparisons on the production side and no satisfactory solution is available There therefore appear to be advantages and disadvantages t o both the production and expenditure approaches The production approach has the merit OF basing its price information directly on the producer price concept This is in contrast to the expenditure approach, where a n u m b e r of adjustments are required, potentially introducing substantial measurement errors. The production approach is /35/ OECD Economic Studies No. 27, /996/11 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA also the only approach that allows the derivation of price information for intermediate goods. However, for investment goods the production approach tends to offer less information. In principle, detailed productivity comparisons might benefit from a mix of the two approaches. Therefore, this paper uses a mix of industry-of-origin and expenditure PPPs for the conversion of value added to a common currency. Table AI shows the conversion factors that were used by industry. Industry-of-origin price comparisons for the nine countries (see source note to Table A2) were used where possible, and could be applied in more than 65 per cent of all cases These price ratios are all based on binary price comparisons between the United States and one other country j 4 Conversion factors from the McKinsey studies were used for Japan and Germany for those industries where these estimates were availabie (Gersbach and Van Ark, 1995) Expenditure PPPs, adjusted for net indirect tax rates and industry-specific distribution margins, were used for the other cases.I5 Nevertheless, for about 20 cases n o suitable PPP was available from either the production or expenditure side, or the basic data were inadequate, impiying that no productivity level could be estimated Table AI indicates that for both the European countries and Japan, 1987 manufacturing price levels were on average substantially above US price levels. T h e relative price levels of Canada and Australia were - on average - almost identical to those in the United States. The variation in price levels across industries is considerable, however, particularly in Japan, and to a lesser degree also in France and the Netherlands. Following the derivation of the PPPs, the main problem for the estimation of productivity levels is the availability of a suitable and comparable data-base. The starting point for data collection was OECD’s STAN database {OECD, 1995a), and the industry detail presented in that database. However, these data are closely linked to the national accounts of each country, which often implies that output and employment information are not derived from consistent data sources (Van Ark, forthcoming). In addition, since STAN is an estimated data-base, some of the industry data appeared to be implausible when compared across countries. For most countries, detailed information was therefore derived from national production censuses These sources have the advantage that output and employment information are based on a single source.I6 Furthermore, the industry detail available in production censuses often allows the reclassification of industries to achieve cross-country comparability. An adjustment for hours worked was based on the estimates of hours worked in the country-specific studies (see note to Table A2). /136 Relative labour productivity was subsequently calculated on the basis of the conversion factors of Table A I , and the value added, employment and h o u r s worked information from the production censuses and national studies. As shown in Table A2, the results indicate a wide range in labour productivity levels. Table A l . Relative price levels for manufacturing industries, major OECD economies, I987 Industry-specific Purchasing Power Parity divided by the Exchange Rate, United States = 100 Industrial sectors (STAN classification) Food Products Beverages Tobacco Textiles Clothing Leather products Footwear Wood products Furniture Paper products Printing, publishing Industrial chemicals Drugs and medicines Chemical products, nec Petroleum refineries Petroleum and coal products Rubber products Plastic products Pottery, china, etc Glass products Non-meta I 1 i c m i nera I products Iron and steel Non-ferrous metals Metal products Office and computing machinery Machinery and equipment, nec Radio, TV and comm equipment Electrical apparatus, nec Shipbuilding and repair Railroad equipment Motor vehicles Motorcycles and bicycles Aircraft i lapan Germany France United Kingdom Canada I84 0 153 I 78 3 I25 6 I23 9 I44 4 I44 4 326 0 115.1 132.4 67.2 145.0 161.7 123.6 156.1 149.4 123.I 145.4 118.0 169.2 1 i 1.6 1 11.6 107.8 130 I 96 1 77 8 11 9 13 2 94 I 94 1 50 5 1128 134 1 74 7 106 5 105 7 91 6 92 2 I06 0 390.3 188.4 197.0 54.4 1300 125.6 124.1 171 I 171.7 235.6 161.9 105.7 184 7 145 2 145 2 73 9 73 9 86 6 84 7 30 9 30 9 30 9 00 7 60 8 96 8 06 0 01 7 96 0 102 7 142.4 122.4 122.4 09.4 09.4 28.9 42.4 10.6 35.9 10.6 04.4 22.7 26.7 27.2 36.7 162 8 148.3 139.9 139.9 139.9 120. I 109.5 106.7 149.4 22 I .8 159.0 197.2 255.0 103 6 I03 6 I03 6 105 4 105 4 89 9 89 9 I06 2 106 2 106 2 103 9 121 7 109 3 99 8 99 8 120 9 120 9 82 9 112.8 89.0 123.6 132.0 100 0 92.8 101 3 124. I 97 3 97 3 97 3 101 4 101 4 92 9 97 3 99 0 99 0 99 0 97 1 108 1 98 7 105.1 105.1 I24 5 120.6 95.4 243.5 92 I 2 12.2 226.5 187.6 167.9 n.a. na na na na 1128 na 191.9 121.9 125.5 179.7 94.4 n.a. 97.5 139.9 95.0 95.0 95.0 125.1 n.a. 144.5 118.6 118.6 136.4 162.9 Australia Netherlands Sweden 93 4 120 4 I53 8 85 5 108 1 I52 8 47 6 64 2 67 2 1190 14 5 166 1 1105 24 4 153 3 127 3zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJ 96 2 132 3 90 9 96 2 132 3 144 7 36 7 160 1 115.4 198.1 126 1 1130 1129 94.0 250.2 101 7 101 7 188.5 93 1 93 1 93 1 92.4 na 87 8 93 1 103 8 103 8 103 8 I04 9 1 I96 110 1 75.8 75.8 131 5 142 7 149. I 141.0 79 7 101 7 Ill I Ill 1 101 7 101 7 91 3 91 3 91 3 I42 6 na 139.8 163.2 163.2 126.0 139.3 166.4 131.7 122 4 122 4 122 4 I28 2 128 2 103 0 122 4 1353 135 3 135 3 111 2 I30 0 1125 142.9 142.9 138 2 134 I 2 10.6 266.0 122 4 na 227.8 Economic Studies No. 2 7, I996111 .-..-zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA CT, CI) zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA h 3 \ U C 2 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA L 3 -m U C - c zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 0 ..-0zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA U c-' m C a, .-m L m > U Lc- E U m a, 0 C a, .U m E 2 a, U 0 X zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA M C W U Table A2 Manufacturing labour productivity levels in major zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA OECD economies, I987 Value added per hour worked, leader country = 100' ~ Industrial sectors (STAN classification) Food products Beverages Tobacco Texti 1es Clothing Leather products Footwear Wood products Furniture Paper products Printing , publishing lndust ria I chemicals Drugs and medicines Chemical products, nec Petroleum refineries Petroleum and coal products Rubber products Plastic products Pottery, china, etc. Glass products Non-metallic mineral products Iron and steel Non-ferrous metals Metal products Office and computing machinery Machinery and equipment,nec Radio, TV and comm. equipment Electrical apparatus, nec Shipbuilding and repair Ra i I road equipment Motor vehicles Motorcycles and bicycles Aircraft ~~ United States lapan Germany France United Kingdom 100.0 100.0 75 0 64 0 90 I 74 9 70 7 59 6 88 I 87 8 100.0 86 4 100.0 31 7 47 6 38 1 41 0 41 5 37 3 51 8 14 I I8 9 42 3 47 5 56 6 83 0 66 1 62 9 41 8 96 7 42 1 39 7 81 5 48 5 100.0 60 1 86 5 52 9 92 6 84 8 80 2 100.0 52 5 95 I 52 8 41 7 66 3 54 6 65 3 61 6 68 4 71 9 55 0 49 4 57 0 57 1 53 4 54 5 61 2 58 5 61 2 54 2 78 4 61 2 66 7 57 4 66 5 69 3 79 5 86 8 62 6 78 7 79 5 59 4 60 9 33 5 71 5 42 5 57 6 66 8 53 5 100.0 69 4 61 5 63 8 68 6 60 0 46 3 56 0 67 3 60 8 52 3 54 4 30 6 na 86 5 60 9 71 I 93 9 100.0 54 6 na 65 3 43 6 64 6 45 1 52 3 51 0 52 6 61 3 33 7 47 4 30 0 87 6 61 2 59 8 58 7 73 8 46 6 72 7 69 7 43 6 51 6 65 4 68 5 61 0 57 6 71 6 67 9 53 5 50 6 36 4 27 5 51 5 100.0 57 7 77 3 88 4 75 0 69 7 79 4 67 4 83 9 94 5 98 2 75 8 100.0 100.0 100.0 74 2 90 7 I00.0 100.0 IOO.0 100.0 93 5 86 7 64 5 28 6 100.0 83 5 Canada 62 7 44 0 65 6 64 2 63 5 55 1 62 2 58 9 67 4 82 3 60 3 90 1 73 5 65 2 44 0 Australia 47 7 49 0 42 2 44 0 47 7 48 5 51 2 27 2 43 5 41 7 59 5 57 8 38 9 47 I 24 5 100.0 na 77 4 67 2 97 5 63 8 68 0 72 4 100.8 79 8 42 1 70 2 71 4 61 2 49 1 42 9 69 7 57 5 54 5 45 5 68 2 47 8 40 8 74 9 48 1 33 9 67 1 46 6 36 8 32 3 23 3 44 7 Netherlands 83 3 76 5 63 4 100.0 100.0 100.0 104.0 100.0 100.0 85 9 51 0 100.0 72 7 89 8 86 6 76 0 100.0 100.0 100.0 100.0 86 9 59 6 na 78 4 35 8 65 7 91 4 87 7 51 7 na 53 2 Sweden 62 2 58 3 68 7 62 1 64 6 78 I 62 I 54 4 80 4 100.0 70 1 65 7 75 3 52 9 100.0 89 8 86 6 75 5 58 9 80 0 69 6 80 3 93 5 100.0 40 7 74 1 78 4 72 3 39 0 20 8 66 6 na na na na na na 78 3 29 6 56 2 33 5 33 7 38 8 OECD Economic Studies No. 2 7, /996/11 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA C W U W 3 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA v) in U -cm L W 5 W 2 .-m E r in z m -o m C m “a O“ 09 - d Y., - N zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA U 5 U M .-c Y U -01 odM MY., -? o\ Y., - 6 N W .- 4.a C 3 U W U?-’ C - M E Ll. NY., 0 CO - 9 M > C m 2 U C m a m - b.01 L” - N M d ‘0 O -b. d U) W r m 65 7J W 4-2 .- C 3 -c 0 E ‘mr; 9u .U c 25 U) - m .m 7? L z3 z 4 -ofj140 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA f’” Competition, Productivity and Efjkiency zyxwvuts NOTES Efficiency and productivity are related, but not identical concepts (Sharpe, 1995). A firm I. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA o r industry is considered t o be inefficient if it could produce more output with existing inputs, be. the firm is not on the production possibility curve, but within it. Productivity relates the quantity of output produced to one o r more inputs used in its production, irrespective of the efficiency of their use. 2. The paper by Symeonides (I997) provides an extensive discussion of the link between competition and innovation. 3. Finland’s rapid productivity gains over the past decade are closely related to the significant restructuring of i t s manufacturing sector over the period I99 I - 1994. 4. Table I shows only estimates of labour productivity. However, the high share of labour compensation in total value added implies that labour productivity levels tend to be a reasonable approximation of TFP levels. TFP levels are more difficult t o calculate as the measurement of real capital stocks across countries poses several methodological difficulties (Blades, 1993; Maddison, 1993). Table 5 below presents some estimates of TFP levels for total manufacturing, based on standardised estimates of capital stocks for these countries (see Van Ark, forthcoming). 5. T o convert value added t o a common currency, PPPs from available industry-of-origin studies were applied where possible, whereas (adjusted) expenditwe PPPs were applied in the other cases (see Annex for methodological details). For a few industries no suitable PPP could be estimated, or the basis data were inadequate, and consequently no productivity estimate could be derived. 6. The productivity estimates in Table 2 differ somewhat from those in the countryspecific studies because of two main methodological differences. First, the countryspecific studies do not estimate conversion factors for all industries, but sometimes apply the PPP of one industry to another industry. In most cases, the current study applies expenditure-based PPPs to industries for which no industry-of-origin estimate is available. Second, the industry breakdown in the current study is more detailed than that of the country-specific studies, which affects the estimated productivity level for total manufacturing. The Annex and Pilat (I996) provide more details on the estimation procedure. 7. Table 2 presents results only for the largest sectors. More detailed estimates, for 35 manufacturing industries in 9 countries, are available in the Annex. 141/ OECD Economic Studies No. 2 7, I996111 8. The productivity estimates in Table 3 differ somewhat from those in Table 2 and Table A2, primarily due t o differences in the basic data, related t o the precise definition of an industry. However, the main thrust of the results tends to be the same. 9. A disadvantage of these studies is their rather historical character, as most of them cover inefficiency in the 1970s. 10. In some services (e.g. electricity and transport), physical measures are a relatively sound basis t o measure real output, thus reducing problems with the conversion of output to a common currency. For some other services (e.g. distribution, construction) comparisons of expenditure price levels provide a reasonable basis to convert real output (e.g. Baily, 1993), as prices of these services tend t o be influenced only little by trade and transport margins o r international trade (see Annex). 11. Average levels of capital productivity and capital intensity may also conceal a large variation in capital intensity and capital productivity across countries. For instance, the low level of capital intensity in Japanese manufacturing obscures large differences among industries (Pilat, I996), with the iron and steel industry, shipbuilding and motor vehicles in Japan having very high levels of capital intensity compared t o these industries in other OECD countries. 12. This factor is related t o X-inefficiency (Leibenstein, I966), which are efficiency differences resulting from organisational, effort and skill-related factors. ( I 996). 13. A more extensive description of data and methodology is available in Pi4at zyxwvutsrqponmlkjihgfe Most industry-of-origin studies have taken the United States as the reference basis for 14. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA productivity comparisons, partly because the United States is generally considered t o be the productivity leader in manufacturing, and partly because the quality of its data. There are also studies (O’Mahony, 1992; Freudenberg and Unal-Kesenci, 1994) that have taken a European country as the basis, e.g. Germany or the United Kingdom. The results from these various studies are generally not transi-tive, i.e. a comparison between t w o countries may not be consistent with a comparison through a third country. This issue is not addressed here, although recent studies have shown how transitive results can be derived (Pilat and Prasada Rao, 1996). IS. The shaded figures in Table A I are based o n (adjusted) expenditure PPPs. More detail on the adjustment factors, including the detailed distribution margins by industry, is available in Pilat (I 996). 16. The production census data are discussed in more detail in Van Ark (forthcoming) and in the country-specific studies quoted in that study. Competition, Productivity and Efficiency BIBLlOGRAPHY ABRAMOVITZ, M. ( I 989), Thinking About Growth, Cambridge University Press, Cambridge. ARK, B. Van (I993), International Comparisons of Output and Productivity, Groningen Growth and Development Centre, Monograph Series, No. I, Groningen. ARK, B. Van (I995), “Produccion y productividad en el sector manufacturer0 espafiol. Un analisis comparativo I950- I 992”, lnformacion Comercial Espaiiola. La actividad empreserial en Espaiia, No. 746, pp. 67-77, Madrid. ARK, B. Van ( I996), “Productivity and competitiveness in manufacturing: A comparison of Europe, Japan and the United States”, in: B. Van Ark and K. Wagner (eds.), International Productivity Differences: Measurement and Explanations, North Holland, Amszerdam, pp. 23-52. ARK, B. Van (forthcoming), The Economics of Convergence - A Comparative Analysis of Industrial zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Edward Elgar, Aldershot. Productivity since I 950, ARK, B. Van and E. MONNIKHOF (I996), “Size distribution of output and employment: A data set for manufacturing industries in five OECD countries, 1960s- I990”, OECD Economics Department Working Paper, No. 166, Paris. ARK, B. Van, and D. PILAT (I993), “Productivity Levels in Germany, Japan and the United States: differences and causes”, Brookings Papers on Economic Activity: Microeconomics, N o 2, December, pp. 1-48. ARK, B. Van and K. WAGNER (eds.) ( I996), International Productivity Differences: Measurement and Explanations, North Holland, Amsterdam. BAILY, M.N. ( l993), “Competition, regulation and efficiency in service industries”, Brookings Papers on Economic Activity: Microeconomics, No. 2, December, pp. 7 I - 130. BAILY, M. and H. GERSBACH (I 995), “Efficiency in manufacturing and the need for global competition”, Brookings Papers on Economic Activity: Microeconomics, pp. 307-358. BLADES, D. (I993), “Comparing capital stocks”, in: A. Szirmai, B. Van Ark and D.Pilae (eds.), Explaining Economic Growth, North Holland, Amsterdam, pp. 399-409. Bureau of Labor Statistics (BLS) (I995), “Output per Hour, Hourly Compensation, and Unit Labor Costs in Manufacturing - Indexes, 1950- I993”, Office of Productivity and Technology, Washington, DC. CAVES, R.E. et al. (I 992), Industrial Efficiency in Six Nations, MIT Press, Cambridge, MA. CAVES, R.E. and D.R. BARTON (1990), Efficiency in U.S. Manufacturing Industries, MIT Press, Cambridge. 143/ OECD Economic Studies No. 2 7, I996I11 COE, D.T. and E. HELPMAN ( I 995), “International R&D spillovers”, European Economic Review 39, pp. 859-887. EATON, J. AND S. KORTUM ( I 995a), “Trade in ideas: patenting and productivity in the OECD”, NBER Working Paper No. 5049, Cambridge MA. EATON, J. and S. KORTUM ( I 995b), “Engines of growth: domestic and foreign sources of innovation”, NBER Working Paper No. 5207, Cambridge MA. ENGLANDER, S. and A. GURNEY ( I 994), “OECD productivity growth: medium-term trends”, OECD Economic Studies, No. 22, Spring, pp. I I I- 129. ENGLANDER, S. and A. GURNEY ( I 994a), “Medium-term determinants of OECD productivity”, OECD Economic Studies, No. 22, Spring, pp. 49- 109. European Commission (EC) ( I993), Retailing in the European Single Market 1993, Brussek. FECHER, F. and S. PERELMAN ( I992), “Productivity growth and technical efficiency in OECD industrial activities”, in Caves ( I 992), pp. 459-488. FREUDENBERG, M. and D. UNAL-KESENCI ( I 994), “France-Allemagne : prix et productivite dans le secteur manufacturier”, €conomie Internationale, No. 60, 4e trimestre I 994, pp. 33-70. GERSBACH, H. and B. VAN ARK ( I 995), “Micro foundations for international productivity comparisons”, Research Memorandum, No. 572, Institute of Economic Research, University of Groningen. HARRIS, R.G., M.CAVE, T. NAMBU, H. TSUBOUCHI and F. KUROSAWA ( I 995), “International comparison of privatization and deregulation among the USA, the UK and Japan - Volume I: Telecommunications”, The Keizai Bunseki, No. I4 I , December, Tokyo. HASKEL, J. ( I 99 I), “Imperfect competition, work practices and productivity growth”, Oxford Bulletin of Economics and Statistics, Vol. 53, No. 3, pp. 265-279. HOOPER, P. and E. VRANKOVICH ( 1995), “International comparisons of the levels of unit labor costs in manufacturing”, International Finance Discussion Papers, No. 527, Board of Governors of the Federal Reserve System, Washington D.C. H0J,J., T. KATO and D. PILAT ( I 996), “Deregulation and privatisation in the service sector”, OECD Economic Studies, No. 25, 1995/11, pp. 37-74, Paris. JORGENSON, D.W. and M. KURODA ( I992), “Productivity and international competitiveness in Japan and the United States, 1960-1985 ”, Economic Studies Quarterly, Vol. 43, December, pp. 3 13-325. KLEIJWEG, A.J.M. and M.H.C. LEVER (I994), “Entry and exit in Dutch manufacturing industries”, EIM Research Report, No. 9409. KURODA, M. ( I 996), “International competitiveness and Japanese industries”, in B. Van Ark and K. Wagner (eds.), International Productivity Differences: Measurement and Explanations, North Holland, Amsterdam, pp. 107- I4 I. LEIBENSTEIN, H. ( I966), “Allocative efficiency vs. X-efficiency”, American Economic Review, Vol. 56, pp. 392-415. MADDISON, A. ( I 993), “5tandardised estimates of fixed capital stocks: a six country comparison”, lnnovazione e Materie Prime, April. Competition, Productivity and Eficiency zyxwvuts Management and Coordination Agency (I989/90 and I993/94), japan Statistical Yearbook, Statistics Bureau, Tokyo. MAYES, D., C. HARRIS and M. LANSBURY (1994), lnefficiency in Industry, Harvester Wheatsheaf, New York. McKinsey ( I 992), Service Sector Productivity, McKinsey Global Institute, Washington D.C. McKinsey ( I 993), Manufacturing Productivity, McKinsey Global Institute, Washington D.C. McKinsey ( I 994), Employment Performance, McKinsey Global Institute, Washington D.C. McKinsey ( I995), Sweden’s Economic Performance, McKinsey Global Institute, Stockholm. NICKELL, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA S.J. ( I 996), “Competition and Corporate Performance”, journal of Political Economy, Vol. 104, NO. 4, pp. 724-746. OECD ( I987), Purchasing Power Parities and Real Expenditures, 1985, Paris. OECD ( I990), Competition Policy and the Deregulation of Road Transport, Paris. OECD ( I993), Purchasing Power Parities and Real Expenditures, 1990, Vol. 2, GK Results, Paris. OECD ( I994), Assessing Structural Reform: Lessons for the future, Paris. OECD ( I995), “Structural Change and Employment: Empirical Evidence for 8 OECD Countries, STI Review, No. 15, pp. 133-175. OECD ( I995a), The OECD STAN Database for Industrial Analysis, 1974- 1993, Paris. OECD (I995b), Communications Outlook, Paris. OECD ( I995c), “Regulatory reform, market access and international contestability of markets - analytical framework”, TD/TCNVP(95)87, Paris. OECD (I995d), The OECD ANBERD Datbase, Paris. OECD ( I996), Technology, Productivity and job Creation, Paris. OECD (I996a), Purchasing Power Parities and Real Expenditures, 1993, Vol. I, EKS Results, Paris. OECD ( I996b), Indicators of Government Assistance, Paris. OECDllEA (I995), Electricity Information - 1994, Paris. O’MAHONY ( I992), “Productivity levels in British and German manufacturing industry”, National Institute Economic Review, Number 139, February, pp. 46-63. OLIVEIRA MARTINS, I., S. SCARPETTA and D. PILAT (I997), “Mark-up pricing, market structure and the business cycle”, OECD Economic Studies, No. 27, Vol. 1996/11, OECD, Paris. PAIGE, D. and G. BOMBACH (I959), A Comparison of National Output and Productivity, OEEC, Paris. PERA, A. (I989), “Deregulation and privatization in an economy-wide context”, OECD Economic Studies, No. 12, Spring, pp. 159-204. PERELMAN, S. (I995), “R&D, technological progress and efficiency change in industrial activities”, Review of Income and Wealth, 4 I, pp. 349-366. PERELMAN, S. and P. PESTIEAU (1994), “A Comparative performance study of postal services: A productive efficiency approach”, Annales d’€conomie et de Sfatistique, No. 33, pp. 187-202. 4 OECD Economic Studies No. 2 7, I996111 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA PERES LOPES, L.M. (I 994), “Manufacturing productivity in Portugal in a comparative perspec- tive”, Notas Economicas, No. 4, Universidade de Coimbra. PESTIEAU, P. ( I993), “Performance and competition in services”, in Market Services and European Integration, European Economy, No. 3, pp. 125- 148. PILAT, D. ( I996), “Labour productivity levels in OECD Countries: estimates for manufacturing and selected service sectors”, OECD Economics Department Working Papers, No. 169, OECD, Paris. PILAT, D. and D.S. PRASADA R A 0 (I 996), “Multilateral comparisons of output, productivity and purchasing power parities in manufacturing”, Review of Income and Weakh, Vol. 42, NO. 2, pp. I 13- 130. SALTER, W.E.G. ( I966), Productivity and Technical Change, Cambridge University Press, Cambridge. SCHWALBACH, J. ( I99 I), “Entry, Exit Concentration and Market Contestability”, in P. Geroski and J. Schwalbach (eds.), Entry and Market Contestability, Basil Blackwell, Oxford. SHARPE, A. ( I 995), “International perspectives on productivity and efficiency”, Review of Income and Wealth, Vol. 41, No. 2, June, pp. 221-237. STEEDMAN, H. and K-WAGNER (I 989), “Productivity, machinery and skills: clothing manufacture in Britain and Germany”, National Institute Economic Review, No. 128. SYMEONIDES, G. (I 997), “Innovation, firm size and market structure: Schumpeterian Hypotheses and Some New Themes”, OECD Economic Studies, No. 27, Vol. 1996111, OECD, Paris (forthcoming). WINSTON, C. ( I993), “Economic deregulation: days of reckoning for microeconomists”, journal of Economic Literature, Vol. XXXI, September, pp. 1263- 1289.