Productivity and transition in Swedish iron and steel, 1870–1940

Abstract

This paper presents a long-run analysis of an industrial branch in Europe's periphery. It examines how the Swedish iron and steel industry reacted to the strains of increasing competition on world markets which affected the branch between 1870 and 1940. The first part of the paper presents a breakdown into periods. We then go on to analyse the sources of growth with both a primal and a dual approach and look at the dynamics between factor substitution and relative prices. Finally, we examine the contribution of the different factors during the periods of transition. Overall we find total factor productivity as the main responsible force for overcoming the effects of competitive pressure. Some of this aggregate technological change can also be traced to a more intensive use of inputs and to extensive growth, indicative of structural change in the industry. Throughout the period examined the industry reacted to increased competition by process and organisational transformation.

Keywords:

• crisis
• iron and steel
• Sweden
• total factor productivity
• dual estimate

JEL classification:

• D24
• L61
• N13
• N14
• N63
• N64
• O47

Acknowledgements

We thank Lennart Schön, Carl-Axel Nilson and Joan Rosés for help, observations, stimulus and suggestions. Useful comments have been received at presentations held at Universidad Carlos III de Madrid, Lund University and the 7th European Historical Economics Society Conference. Portions of the research were supported by Spanish Ministerio de Educación y Ciencia, projects ECO2009-13331-C01-01, and PR2005-0062. The usual disclaimer applies to the errors that may remain.

Notes

¹Boëthius, ‘Swedish Iron and Steel’ (1958), 171.

²Tortella, ‘Sweden and Spain’ (2005).

³Schön, ‘Sweden’ (2008), 3.

⁴Schön, ‘Technological Waves’ (2009).

⁵Fritz, ‘Swedish Industrialisation’ (2005), 149. Schmitz, ‘What Determines Productivity?’, (2005).

⁶We concentrate on the era of steel. For a thorough discussion of the previous century see Rydén, ‘Skill and Technical Change’ (1998).

⁷As ironmasters became aware of their vulnerability in international markets – stiffer competition with British and Russian producers – their attitudes towards production gradually changed. They came to realize that they had to improve the ironwork's productivity and efficiency to produce a high-quality iron at a low cost. Rydén, ‘Skill and Technical Change’ (1998), 393.

⁸Montgomery, Rise (1939), 84–5.

⁹See Rydén, ‘Skill and Technical Change’ (1998), 395 and 404–5 for more detailed descriptions.

¹⁰Wengenroth, Unternehmensstrategien (1986).

¹¹Montgomery, Rise (1939), 168, 173.

¹²Söderlund, ‘Swedish Iron Industry’ (1958), 54–5, 67.

¹³Söderlund/Wretblad, Fagerstabrukens Historia (1957), 1–17.

¹⁴Boëthius, ‘Swedish Iron and Steel’ (1958), 169–70.

¹⁵Montgomery, Rise (1939), 244.

¹⁶We leave aside a prior transition because we lack the necessary data for contrast. This previous transition took place at the beginning of the nineteenth century with the spread of puddled coke iron in Great Britain and initiated a period of change between 1830 and 1870. It shifted from more traditional processes to the Lancashire method in Sweden as a consequence. ‘Heckscher once posed the question whether the rescue of the Swedish iron industry at the beginning of the nineteenth century, in face of the devastating competition of puddled iron, did not ‘constitute the most glorious page in the history of the iron industry and even the whole economic history of Sweden’. Boëthius, ‘Swedish Iron and Steel’ (1958), 170–71. Also Montgomery, Rise (1939), 79:

‘[I]t was only towards the end of the eighteenth century that the puddling process was invented and brought such perfection as definitely to revolutionize the trade. Henceforth, it became possible to use pit coal in the manufacture of malleable iron, and the Swedish export trade was soon exposed to fierce competition.’

And Heckscher/Söderlund, ‘Rise of Industry’ (1953), 50–1.

¹⁷‘Robert Hadfield's discovery of manganese steel in 1882 is generally recognized as the landmark in metallurgical history that inaugurated the age of alloy steels.’ Tweedale, ‘Sir Robert Abbott Hadfield F.R.S.’ (1985), 63. Invar was invented in 1889. In 1895, Hans Goldschmidt developed the aluminothermic reduction process for producing carbon-free chromium. In 1911, P. Monnartz and W. Borchers discovered the correlation between chromium content and corrosion resistance and also published detailed works on the effects of molybdenum on corrosion resistance. On 13 August 1913, Harry Brearley created a steel with 12.8% chromium and 0.24% carbon, argued to be the first ever stainless steel.

¹⁸Attman, Fagerstabrukens Historia (1958), 641–42.

¹⁹Thereby we capture the total volume of iron and steel product as opposed to a final product approach in which we would have to aggregate the final products of the iron and steel industry with those of metal transformation sectors. This will introduce bias as the metal working implemented quality improvements via technological improvements in the final transformation, but this approach has the advantage of being parsimonious and allows us to discriminate price and volume changes. For a discussion on statistical aggregation for the industry see Shone, ‘Source and Nature’ (1950), 464–5.

²⁰Export statistics are a very reliable source of volume data because information is complete and uniform.

²¹For 1890–1940, the data for cast iron and rough iron production and commercial forged and rolled products were taken from Wistrand, Järnverksföreningen (1938), 53–4, and the Swedish Official Statistics [SOS]. For 1870–1889, cast iron and rough bars were represented by cast iron and rolled and forged iron and steel bars by puddled bars. Both series were taken from the SOS.

²²Cast iron becomes such a small percentage of production that we exclude it.

²³In the TFP calculations to follow, we have multiplied the volume sub-aggregates of each series by its price series.

²⁴Jörberg, Growth and Fluctuations (1961), 68.

²⁵Nilsson, Järn och stål (1972), 161, calculates that home markets grew three-fold between 1885 and 1912 – at a much higher rate than export markets.

²⁶Jörberg, Growth and Fluctuations (1961), 71.

³⁰Söderlund, ‘Swedish Iron Industry’ (1958), 77.

²⁷Direct measurement is very difficult, mainly because we lack survey, censuses and capital-market transactions between owners and users, but also because capital is a produced means of production and durable. See Hulton, ‘Measurement of capital’ (1990), 121–2.

²⁸Gårlund, Svensk industrifinansiering (1947).

²⁹Uddeholms, Forsbacka Jernverks, Fagersta Bruks, Hellefors Bruks, Sandvikens Jernverks and Horndals Jernverks.

³¹Holmquist, Kapitalbildning (2003), 145–46, gives an estimate for capital stock in the metal-working industry based on fire insurance records and horse power, which was used as a control variable.

³²We have extrapolated the values backwards linearly from 1893 to 1870 assuming that installed horse power doubled over that period. We have depreciated the horse power series assuming a 50-year-machinery life with a perpetual inventory method. Scrapping was performed on a FIFO basis. The missing initial horse power stock was calculated following a method similar to Harberger, ‘Primer’ (1978).

³³Labour at rolling mills had 61,5 hours/week, two shifts would total 125 hours. To attain the same amount of labour after the 1919 legislation three shifts of 48 hours total 144 hours were needed. Söderlund, ‘Swedish Iron 1932–1939’ (1959), 48.

³⁴Jörberg (1961), Montgomery (1939), Heckscher & Söderlund (1953).

³⁵Montgomery, Rise (1939), 171.

³⁶Smith, Industry in Sweden (1953), 154–5.

³⁷The first coke furnace mill had not been completed before 1917 in Oxelösunds Järnverks.

³⁸Smith, Industry in Sweden (1953), 155. Output per day in charcoal furnaces at the beginning of the 1950s was 40 tons a day compared to 14 tons around 1900. Only two plants produced more than 100 tons a day. Ten coke furnaces produced an average of 160 tons a day in 1948.

³⁹Lindahl et al., National Income (1930), 200.

⁴⁰Arpi, Den svenska (1951).

⁴¹Openshaw, ‘Measuring charcoal’ (1983) assumes a weight of 115 kg of charcoal per cubic meter of charcoal made from pine wood. Calculations have been repeated with Wibe, ‘Svensk järnhantering’ (1980), 411, who assumes an energy equivalence of 58.82 Hl of charcoal to a ton of coke, with no notable differences.

⁴²Sweden had very small coal deposits and we therefore assume that all coal and coke consumed was imported.

⁴³Söderlund, ‘Swedish Iron, 1932–1939’ (1959), 7n.

⁴⁴Söderlund, ‘Swedish Iron Industry’ (1958), 65–66. Söderlund ‘Swedish Iron, 1932–1939’ (1959), 51–3.

⁴⁵Söderlund/Wretblad, Fagerstabrukens Historia (1957), 119.

⁴⁶Given we have no way of measuring scrap before 1919 we have assigned a higher iron content (55%) to the approximated amount of iron ore consumed, thereby biasing the metallic iron input upwards.

⁴⁷Taken from the Svensk sparbanktidskrift (1934), 825 ff., Bagge, Wages (1935), 259–60.

⁴⁸The rental rate of capital is equal to interest rate plus depreciation. If the depreciation rate is assumed constant and depreciation linear, the part in capital rental cost which reflects changes in marginal productivity is interest rate.

⁴⁹Jungenfeldt, Löneandelen (1966), 241–43.

⁵⁰Huberman, ‘Working Hours’ (2002), 15.

⁵¹Jörberg, History of Prices (1972), 565–66.

⁵²Nilsson, Järn och stål (1972), Table 1: 1–13, 153; Table 1: 10–13, 154; Table 3: 1–13, 165 and Table 3: 10–13, 166. Ljungberg Deflatorer (1988), 24, 55, 82, 280–81. Mitchell, British Historical Statistics (1971), 493–4. Jörberg, History of Prices (1972), 569–70: 16 Värmland, 17 Örebro, 18 Västmanland, and 575-77: 17 Örebro, 18 Västmanland, 19 Kopparberg, 20a Gästrikland.

⁵³Allen, ‘International Competition’ (1979), 920. The assumption that shares remained stable over the whole period is sustainable both because it approximates shares of the iron and steel industry in the US – leading iron and steel industry towards which Sweden could converge – and because the differences with the cost structure of integrated mills vary only slightly. The only major structural change the industry suffers over the period examined is integration and concentration. The cost structure of integrated mills differed very little to that of the industry as a whole in 1913. A further confirmation is that as Swedish industry switches over to the standard European–US blueprint production scheme, i.e. coke pig iron and integrated large mills, their average product of metallic inputs converges to the average of 0.8 which Allen found for three major steel producing countries – Germany, USA and Great Britain.

⁵⁴Profitability and productivity increase, year by year, with little change in organisation, layout and products. Lundberg, Produktivitet (1961), 130–1 discussed in Lazonick/Brush, ‘Horndahl Effect’ (1985), 53–7.

⁵⁵Barro/Sala, Economic Growth (1995), 442–3.

⁵⁶According to Hsieh, ‘What Explains’ (2002), the only condition we need is that output value equal factor incomes, no further assumptions about the form of the production function, bias of technological change, relationship between factor prices and their marginal products are necessary.

⁵⁷Production aggregates fell by around 70% between 1920 and 1921, the cost of capital fell by 22%, wages by 33%, charcoal prices dropped by 61% and metal inputs by 39% between 1920 and 1923. Price adjustments to the sudden drop in output was much more gradual and with the exception of charcoal, which was being replaced with imported coal, factor prices were sticky. The fall in the primal estimate is shown with a star in Figure 5 and the value has been interpolated to maintain levels for comparisons after 1921.

⁵⁸Acid steel was used early on to make ball bearings, an important Swedish innovation. SKF was founded in 1907.