The impact of mechanisation on wages and employment: Evidence from the diffusion of steam power

Leonardo Ridolfi, Carla Salvo, Jacob Weisdorf 17 July 2022



Recent advances in technology have sparked a heated debate about the effects of mechanisation on current societies. A recurring concern expressed by many commentators is that machines will take over tasks previously performed by workers, leading to widespread unemployment and growing inequality. These mounting fears are not new; technology has always generated cultural anxiety throughout history (Mokyr et al. 2015). The path of transition, however, may not be disruptive for all workers and industries. Prominent economists like Adam Smith, for instance, have long emphasised the positive effects of technology arguing that mechanisation will generate significant productivity gains in the long run, leading to cheaper goods, increased labour demand, and thus more jobs. On balance, the net impact of mechanisation on workers and wages is thus ambiguous depending on how the displacement and productivity effects end up weighing against each other (Acemoglu and Restrepo 2018, Autor and Salomons 2018, Atack et al. 2019, Frey 2019, Hötte et al 2022).

Understanding the effects of mechanisation entails difficult measurement problems because waves of contemporary mechanisation typically overlap, making it difficult to isolate the effect of a specific technology. In a recent paper (Ridolfi et al. 2022), we seek to address these concerns by turning to a time when technological development was still at its earlier stages and governmental support for workers was poor. In turn, this makes it easier to identify the influence of mechanisation on labour outcomes, since institutional and other confounding factors were likely to play smaller role in past societies.

We consider one of the most significant waves of mechanisation in history: the rise and spread of steam power in 19th century France. This empirical setting provides an ideal testing ground to explore our main hypotheses. French historical statistics are extraordinarily detailed for the time-period in question and allow us to observe the mechanisation process from the early stages of diffusion until maturity. A peculiar feature in this context is that traditional sources of motive power like water and wind played a significant role in French manufacturing. This means that large variations in local patterns of technological adoption can be better understood compared to England, for example, where the coal-steam model dominated. 

To set up our analysis, we build a new dataset consisting of two parts. The first concerns the information reported in the two earliest industrial surveys carried out by the French Bureau of Statistics during the 19th century (Chanut et al. 2000). The two surveys are all-encompassing and include information on wages, employment, and use of motive power. The second group of data – collected from a variety of sources – concerns district-level information encompassing a wide variety of geographic, demographic, and economic factors including access to water power, closeness to coalfields, quality of the workforce, and the presence of banks, all of which help us identify the local initial conditions that influenced early steam-engine adoption.

We use this information to answer two main questions. First, what were the main drivers of steam engine adoption? Here, we test five hypotheses that have loomed large in the literature: (1) the so-called high-wage hypothesis that argues that expensive labour and cheap energy (coal) prompted producers to innovate (Allen 2009, 2011); (2) the resource-availability hypothesis that instead put emphasis on the availability of cheap coal substitutes such as wind and water as potential brakes of innovation (Crouzet 1996); (3) the health-and-knowledge hypothesis that stresses the quality of labour force in terms of nutrition, health status, and knowledge as a push-factor that helped innovation (Mokyr 1990, Squicciarini and Voigtländer 2015, Kelly et al 2014, 2022); (4) the market-force hypothesis that looks at market size and proximity to technological knowledge as important drivers of innovation (Acemoglu and Linn 2004, Franck and Galor 2021); and, finally, (5) the finance-led-growth hypothesis that stresses the role played by financial institutions in fostering innovations (Madsen and Ang 2016, Rousseau and Sylla 2005). Second, we explore the effects of steam engine adoption on labour market outcomes.

To provide answers, we apply a diff-in-diff approach with propensity score matching, comparing industries that remained without steam power (non-treated group) to those that adopted steam power between the two registrations (treated group). The process of transition was characterised by considerable heterogeneity both across space and within sectors. Figure 1 shows the distribution of treated and non-treated areas in France in the case of flour milling – the most common subindustry in the 1840s. The treated districts are marked in dark blue, and the non-treated ones in light blue. White districts are those where flour milling was absent or steam was already installed in the 1840s. 

Figure 1 Treated and non-treated local flour-milling districts

In terms of factors accounting for steam adoption, our findings provide support to the central mechanisms highlighted in the literature. Many and highly paid workers alongside closeness to coalfields significantly raised the likelihood of steam adoption. Lack of water power or inadequate water supplies also encouraged firms to install steam engines. Scientific as well as technical knowledge significantly increased the probability of adopting steam, while basic literacy skills and university knowledge did not propel the diffusion of steam power. Greater market size, a well-developed transport infrastructure, bank presence, and good local health conditions also raised the likelihood of steam engine adoption.

Concerning the effects of steam-engine adoption on labour market outcomes, our findings oppose the fears that past mechanisation were accompanied by significant wage cuts and job losses. After adjusting for selection effects, we observe that steam-adopting industries ended up employing up to 94% more workers than their non-steam-adopting counterparts (Figure 2). We also discern that steam-adopting industries paid wages that were up to 5% higher on average than in non-adopting ones (Figure 3). The positive effect of mechanisation on wages suggests that technical change has widened the wage structure also during the earlier stages of industrial development and not just more recently (Goldin and Katz 1996, 1998). Furthermore, the idea that steam power was labour augmenting (Rousseau 2008) confronts the Habakkuk thesis that labour shortage led to higher wages and ultimately drove labour-saving industrial innovations (Habakkuk 1962, Allen 2009).

Figure 2 Effect of steam engine adoption on male employment

Figure 3 Effect of steam engine adoption on male wages

Given the special importance of water as a source of energy in French manufacturing, we also explore a subsample of water-powered industries in the 1840s, considering the effect that a partial or full transition to steam had on labour outcomes. We refer to the first instance (partial transition) as technical supplementation and to the second one (full transition) as technical substitution or creative destruction. To identify the effects on wages and employment, we compare the steam-adopting industries with those that continued to rely exclusively on water power between the 1840s and the 1860s. Here, our results suggest that technical substitution significantly increased both the number of male workers employed (by 41%) and their average wages (by 14%) while technical supplementation significantly increased the number of male workers (by 91%) but not their average wages. Creative destruction in this sense was better for workers’ compensation, but technical supplementation was better for employment.


Acemoglu, D and J Linn (2004), “Market Size in Innovation: Theory and Evidence from the Pharmaceutical Industry”, Quarterly Journal of Economics 119: 1049-1090.

Acemoglu, D and P Restrepo (2018), “The Race Between Machine and Man: Implications of Technology for Growth, Factor Shares and Employment”, American Economic Review 108: 1488-1542.

Allen, R C (2009), The British industrial revolution in global perspective, Cambridge University Press.

Allen, R C (2011), “Why the industrial revolution was British”, Economic History Review 64: 357-384.

Atack, J, R A Margo and P W Rhode (2019), “’Automation’ of Manufacturing in the Late Nineteenth Century: The Hand and Machine Labor Study”, Journal of Economic Perspectives 33: 51-70.

Autor, D H and A Salemons (2018), “Is automation labor-displacing? Productivity growth, employment, and the labor share”, Brookings Papers on Economic Activity 49: 1-87.

Chanut, J. M, J Heffer, J Mairesse and G Postel-Vinay (2000), L’industrie française au milieu du 19e siecle. Les enquêtes de la Statistique Générale de la France, Paris: EHESS.

Crouzet, F (1996), “France”, in M Teich and R Porter (eds), The industrial revolution in national context: Europe and the USA, Cambridge University Press.

Franck, R and O Galor (2021), “Flowers of evil? Industrialization and long run development”, Journal of Monetary Economics 117: 108-128.

Frey, C B (2019), The technology trap: Capital, labor, and power in the age of automation, Princeton University Press.

Goldin, C and L F Katz (1996), “Technology, skill, and the wage structure: insights from the past”, American Economic Review 86: 252-257.

Goldin, C and L F Katz (1998), “The Origins of Technology-Skill Complementarity”, Quarterly Journal of Economics 113: 693-732.

Habakkuk, H J (1962), American and British Technology in the Nineteenth Century, Cambridge University Press.

Hötte, K, M Somers and A Theodorakopoulos (2022), “The fear of technology-driven unemployment and its empirical base”,, 10 June.

Kelly, M, J Mokyr and C Ó Gráda (2014), “Precocious Albion: a new interpretation of the British industrial revolution”, Annual Review of Economics 6: 363-389.

Kelly, M, J Mokyr and C Ó Gráda (2022), “The mechanics of the Industrial Revolution”, Journal of Political Economy (forthcoming).

Madsen, J and J Ang (2016), “Finance-Led Growth in the OECD since the Nineteenth Century: How Does Financial Development Transmit to Growth?”, Review of Economics and Statistics 98: 552-572.

Mokyr, J (1990), The Lever of Riches: Technological Creativity and Economic Progress, Oxford University Press.

Mokyr, J, C Vickers and N L Ziebarth (2015), “The History of Technological Anxiety and the Future of Economic Growth: Is This Time Different?”, Journal of Economic Perspectives 29: 31-50.

Ridolfi, L, C Salvo and J Weisdorf (2022), "The Impact of Mechanisation on Wages and Employment: Evidence from the Diffusion of Steam Power", CEPR Discussion Paper No. 17389.

Rousseau, P L (2008), “Biased and Unbiased Technological Change”, in The New Palgrave Dictionary of Economics, Palgrave Macmillan.

Rousseau, P L and R Sylla (2005), “Emerging financial markets and early US growth”, Explorations in Economic History 42: 1-26.

Squicciarini, M P and N Voigtländer (2015), “Human Capital and Industrialization: Evidence from the Age of Enlightenment”, Quarterly Journal of Economics 30: 1825-83.



Topics:  Labour markets Productivity and Innovation

Tags:  steam power, mechanisation, manufacturing, employment, wages, France

Assistant Professor of Economic History, University of Siena

PhD candidate in Socio-Economic and Statistical Studies, Sapienza University of Rome

Professor of Economic History, Sapienza University of Rome


CEPR Policy Research