RDP 2018-10: Wage Growth Puzzles and Technology 2. Technology, Bargaining Power and Nominal Wages Growth

This section looks at some of the ways in which new technology, in particular ICT, may be impacting on the relative bargaining power of employers and employees across a broadening range of industries, and hence on wages growth. It then examines recent RBA research work in this area.

2.1 New Technology, Product Markets and Labour Markets: Some General Observations

Two closely related developments that have had a major impact on both labour and product markets in many countries in recent decades are the application of digital technology (e.g. Brynjolfsson and McAfee 2014) and the associated ‘new’ globalisation.[6] Digitisation – the representation of information in bits – has dramatically reduced the cost of storing, transmitting and manipulating data. Complementing and magnifying the digital revolution have been three important technological developments: the internet and the development of high-speed networks and search engines; the widespread dissemination of platform devices – mobile telephones, tablets, etc – that provide consumers, workers and service providers with both connectivity and access to data and information in most places at all times; and the use of digital technology and platforms to collect and merge masses of exploitable commercial, personal and geographic data: so-called big data.

Reference is often made to increased consumer price awareness and sensitivity across many sectors due to growing access to online price comparisons, putting greater pressure on companies to reduce costs – including labour costs (e.g. Akerman, Leuven and Mogstad 2017). But the take-up of ICT is also having a more pervasive impact on many other aspects of business, including marketing, distribution, just-in-time inventory control, automation and the spread of web-based service providers. It has also made it easier and cheaper to coordinate complex activities across different locations, allowing firms to take greater advantage of geographical differences in wages and other costs by ‘unbundling’ production processes through the creation of global supply chains: the new globalisation. A number of studies both in Australia and overseas have highlighted the lengthening of global supply chains.[7]

The ICT revolution and associated new globalisation have enabled and encouraged major changes in work and management practices across many industries, in particular by making both production processes and labour input increasingly ‘divisible’ across countries and firms and within firms. Digital technologies make it easier and cheaper for firms to keep in touch with and monitor employees, whether they are in the office, at home, in different offices within the same country or in different countries. This has enabled firms wanting to increase flexibility and reduce costs to use cheaper overseas workers in parts of their production process and/or non-standard forms of employment domestically, such as self-employed, contract and casual work. While in many cases this has been beneficial for employees looking for more flexible working arrangements that meet their specific circumstances, a number of studies suggest it has also been an important factor behind declining union membership and employee bargaining power and – especially in sectors and occupations where digitisation is extending to automation – falling job security and lower wages.[8]

The question of whether the ICT revolution and associated new globalisation will continue to have a substantial impact on product and labour markets has been at the centre of a lively academic debate in recent years. ‘Techno-pessimists’ such as Gordon (2016) argue that the effect of the ICT revolution is overstated (see also Cowen (2011) and Fernald (2015)). In a major study looking at the rise and fall of productivity growth in the United States, he argues that the impact of the ICT revolution was short-lived and less significant than what he calls the first (steam engine, railroads) and second (electrification, plumbing) industrial revolutions. In his view, the temporary nature of the third (ICT) industrial revolution explains why US productivity growth rose from around the mid 1990s but then fell away after the mid 2000s. As a challenge to those who argue to the contrary, he raises a pertinent question: if the effect of the ICT revolution is really so substantial and ongoing, why is it no longer showing up in faster productivity growth?

On the other side of the argument, so-called ‘techno-optimists’ such as Brynjolfsson and McAfee (2014) see ICT as a ‘general purpose’ technology: pervasive (in the sense of being a potentially important input across many sectors of the economy); subject to continual improvement; and spawning ongoing innovations in many areas. They argue that we are now at an inflexion point, with the impact of this third industrial revolution (or what they call the second machine age) set to become much more widespread as computing power continues to increase exponentially,[9] as more and more activities and devices become digitised and as ongoing innovations such as artificial intelligence, 3D printers and the internet of things are spawned by digitisation.

Increasingly, the evidence seems to suggest that both camps are partially correct. A number of studies have found not only extremely uneven patterns of take-up and utilisation of new technology across countries and within sectors of an economy but also a growing gap between high productivity firms that are at the forefront of adapting and using new technology and laggard firms within the same sector. With reference to a seminal OECD study in this area (Andrews, Criscuolo and Gal 2015), Haldane (2017a, p 3) states the following:

This empirical evidence suggests a long tail of countries and companies with low, slow productivity growth. These productivity laggards have been unable to keep-up, much less catch-up, with frontier countries and companies. At the same time, an upper tail of companies and countries has maintained high and rising levels of productivity. These productivity leaders are pulling ever-further away from the lower tail …

This empirical pattern … helps explain why we might see the co-existence of secular innovation (among leaders) and stagnation (among laggards). It helps account for the fall in productivity growth rates – namely, slower rates of diffusion of new innovation to the long lower tail of companies.

Many of the potential benefits of the ICT revolution to a firm – more detailed information on potential clients, easier marketing, greater flexibility with respect to labour input, greater economies of scale and lower costs through outsourcing and global supply chains – are largely ‘non-rival’ in nature, in the sense that one firm taking advantage of the benefits does not prevent a competitor from doing the same. Nonetheless, the reality is that doing so successfully and over a sustained period of time requires considerable management skill, information and capacity to fund new investment. These requirements are far from uniformly spread across countries or firms. By way of example, a recent study of law firms in Australia found, amongst other things, that all higher profitability firms in the survey are ‘harnessing new technologies’; and that cost and lack of knowledge are key factors inhibiting smaller firms from taking up new technologies (Macquarie Bank 2017).[10] All firms in the survey saw low cost ‘virtual firms’ and outsourced lawyers as the main threats to their business, as clients bypass them as the traditional service providers.

While the take-up of digital technology has been very uneven across firms within a given sector or industry, the evidence suggests that its effects are spreading as it is applied by some firms across more and more sectors, industries and professions (e.g. Weil 2014; McTernan and Reed 2015; Degryse 2016; Katz and Krueger 2016; Goldschmidt and Schmieder 2017). While this is happening faster in the United States than in most other countries, anecdotally the indications are that its impact and application in Australia are also continuing to widen (Department of Industry, Innovation and Science 2016; Productivity Commission 2016). This has been identified clearly in the RBA business liaison program. Lai et al (2018) note the following:

Information from the Bank's liaison program and other survey evidence suggests firms have been more willing to spend on IT than most other forms of capital in recent years. The desire to increase efficiency, lower the cost of doing business, and improve service quality have been important drivers of the adoption of technology across the economy.

The authors also note that the take-up of ICT across firms within a given industry is very uneven, consistent with overseas studies referred to earlier.

While the application of ICT is spreading to more sectors and industries, there is still a good deal of uncertainty as to just which activities, industries and stages of production can or cannot be digitised, automated, unbundled and/or offshored; and with respect to the pace at which new ICT innovations may occur and be taken up. This uncertainty or ‘fear of the unknown’ may itself be having an important impact on labour market behaviour.

2.2 RBA Empirical Work on Technology, Bargaining Power and Nominal Wages Growth

2.2.1 Technology and product markets

Many consequences of the ICT revolution discussed above could in principle help explain why nominal wages growth has been sluggish across most countries, including Australia. Increasing divisibility of labour, falling union membership, outsourcing and increased job insecurity may be shifting bargaining power away from employees and having an important impact on nominal wages growth.

Similarly, greater competition in product markets may have stiffened employer resistance to higher wages and increased employer focus on cutting costs. In a speech in 1996, the then RBA Deputy Governor, Ian Macfarlane, reminded his audience that ‘we now live, work and aim to sell our output in a world … characterised by … low pricing power by business and labour’. He went on to suggest that one of the two most important reasons for low inflation was likely ‘increased competitive pressures as economies have become more open and as domestic competition has intensified’ (Macfarlane 1996, p 18).

While initially increased domestic competition in Australia was typically attributed to market liberalisation measures in the 1980s and 1990s (e.g. Dwyer and Leong 2001), more recently the focus has been on global factors – including globalisation and the use of global supply chains to reduce costs. In a recent speech focused on why nominal wage growth was so low, RBA Governor Lowe suggested that ‘foremost’ among the structural factors at work were perceptions of increased competition:

Many workers feel there is more competition out there, sometimes from workers overseas and sometimes because of advances in technology … [As a consequence] many workers feel like they have less bargaining power than they once did.

Lowe went on to suggest this was not the full story:

It is likely that there is also something happening on the firms' side as well … Businesses are not bidding up wages in the way they might once have. This is partly because business, too, feels the pressure of increased competition.

One response to this competitive pressure is to have a laser-like focus on containing costs … Paying higher wages can sit at odds with that mindset. (Lowe 2017b)

The increasing impact of digital technology is certainly not the only factor that may be leading to increased competition in product markets and greater employer resistance to cost – including wage cost – pressures, but the anecdotal evidence suggests it is an important and growing one. A key rationale for introducing digital technology is to reduce costs. As it is disseminated across more firms and industries, other things equal, competitive pressures increase, including increased competitive pressure on technology laggards from the technology leaders in their sector or industry.

Nonetheless, finding evidence that increased product market competition and greater employer resistance to cost increases are important factors in explaining the short-run wage puzzle is challenging. A key reason for this is the difficulty of finding suitable proxy variables and rich enough data sources to isolate and robustly test the hypothesis when there are so many moving variables. If employer resistance to wage increases was rising then, other things equal, this should be reflected in rising numbers of wage-related industrial disputes, which is not the case (Figure 4). However, other things are not equal: the hypothesis outlined by RBA Governor Lowe in the earlier quotations is that both employer resistance and union bargaining power are being impacted by new technology and putting downward pressure on wages growth. This is more consistent with the industrial disputes data.

Figure 4: Wage-related Industrial Disputes – Australia
Figure 4: Wage-related Industrial Disputes – Australia

Source: ABS

One possible data source that might shed more light on this issue is survey data. Ballantyne and Langcake (2016) used information from the Reserve Bank's business liaison program to look for explanations of a structural break in retail goods inflation around 2010. Amongst their findings were that an intensification of competition and firms' increased efforts to reduce costs along their supply chain were likely contributing factors. In both cases, they suggest, digital technology appeared to have played an important role, both by way of increased consumer awareness of prices via online shopping and increased supply of retailers due to competition from foreign online companies.

It will be argued in Section 4 that the growing divergence in productivity levels between firms in the same industry may well be a key factor behind increasing product market competition, and that a firm-level framework that allows for heterogeneous productivity performance provides a richer framework and database for testing hypotheses in this area.

2.2.2 Technology and labour markets

Most of the RBA research looking to explain surprisingly low nominal wages growth has been done within the framework of a conventional wage Phillips curve relationship:

where w is nominal wages growth, ue the unemployment rate, ue* the NAIRU (the non-accelerating inflation rate of unemployment), z a vector of other exogenous cyclical factors that may affect wages growth and ε the error term. In the RBA's preferred formulation, the vector z includes inflation expectations, actual inflation, trend growth in labour productivity and the change in the unemployment rate.

A recent RBA study (Arsov and Evans 2018), using the formulation in Equation (1), examined why wages growth across a number of advanced economies, including Australia, had surprised on the downside in recent years, despite tight labour markets. It reached the following conclusions. Firstly, Phillips curves would not appear to have flattened over the 2000s: that is, recent low wages growth does not appear to reflect a weaker relationship with unemployment. Secondly, lower inflation expectations and lower productivity growth help explain low wages growth, but only partly, such that ‘over the past two years wages have been persistently weaker than estimated by the models’. The study also noted that this recent period of downward surprises in wages growth is not the first – similar outcomes were seen in the early 2000s.

The above finding – that, even when additional explanatory variables such as low inflation expectations or weak productivity growth are included, conventional Phillips curves cannot fully explain the weakness in wages growth in recent years – has been a consistent theme in recent speeches and in research by the RBA focused on Australian wages growth. It has led the Bank to examine structural as against cyclical factors that may help explain the nominal wage growth surprise. This work is examined in Appendix A.

Three broad observations may be made regarding this RBA research. Firstly, because recent wage growth ‘surprises’ have been surprises relative to what has been forecast by traditional Phillips curve augmented wage equations, it is not surprising that most of the analysis by the RBA and other central banks of possible structural impacts on wage behaviour has been done within a Phillips curve framework. However, while this framework incorporates cyclical shifts in employee bargaining power over the course of the economic cycle through the unemployment gap term, it is arguably less well suited to examining structural factors that may be having a slow-building impact on employee bargaining power over a longer time period. In part, this is because such structural factors are often hard to measure, providing fewer observations and hence making their impact harder to assess within a cyclical Phillips curve framework (Arsov and Evans 2018). More broadly, there is increasing evidence – at least in the United States – of a trend rise in profit mark-ups over recent decades (e.g. Barkai 2016; Bessen 2017; De Loecker and Eckhout 2017; Eggertsson et al 2018). If similar trends are evident in Australia, fully analysing what is going on and incorporating it into both wage and price equations may require more radical surgery. Cockerell and Russell (1995, p 4) noted the following:

Modelling the Australian economy using a price taking model is inconsistent with the observation that firms appear to set prices and that labour market outcomes are the result of collective bargaining between labour and firms (or possibly labour, the government and firms). It may be more appropriate, therefore, to model Australian wage and price inflation within an imperfect competition model.

This point is returned to in Section 4.

The second observation concerns the nature of the structural factors that may be dampening wages growth and that the RBA has examined in its research. They are as follows:[11]

  • increased import competition, globalisation and job insecurity;
  • declining unionisation;
  • casualisation of the labour force; and
  • gradual unwinding of a real wage overhang built up in earlier years.

All of the above factors except the last one seem likely, a priori, to be in part related to the impact of new technology. Moreover, for reasons set out in Appendix A, the gradual unwinding of a real wage overhang seems unconvincing as an explanator of ongoing nominal wage growth surprises.

A third observation relates to the possible link between some of the above factors and the uneven take-up of new technology. The Autor et al (2017) paper notes that superstar firms appear to be ‘increasingly using domestic outsourcing to contracting firms, temporary help agencies, and independent contractors and freelancers for a wider range of activities previously done in-house’ and references some other studies on this point (Weil 2014; Katz and Krueger 2016). The authors go on to state the following:

This fissuring of the workplace can directly reduce the labor share by saving on the wage premia (firm effects) typically paid by large high-wage employers to ordinary workers and by reducing the bargaining power of both in-house and outsourced workers in occupations subject to outsourcing threats and increased labor market competition … (Autor et al 2017, p 26)

A reasonable summary of RBA work examining structural factors that may be reducing employee bargaining power and contributing to lower wages growth is that, while the empirical results help explain some of the weakness, the overall impact is not large and still leaves an unexplained element. This was also the finding of Arsov and Evans (2018) in their examination of the effect of shifts in bargaining power (amongst other factors) in explaining weak wages growth across a number of countries, in which they concluded that:

Structural changes in the labour market, including declines in unionisation rates and employee protections also seem to be factors behind the weaker wage growth. However, all of these factors are only part of the story and, over the past two years, there appears to have been some other common factor weighing on wage growth across advanced economies.

As noted earlier, part of the reason for the limited findings from RBA research in this area may be the absence until recently of a large and rich enough Australian database that, amongst other things, allows for the impact of the explanatory variables being examined to be isolated from other potentially correlated variables. More broadly, finding enough data points (and variations in the data) to be able to robustly test the impact of potential explanatory variables is a common problem across many areas of economic research, and has led some researchers to look for richer, micro-level data sources to formulate and test their hypotheses.

Recently available firm-level data sources in Australia may well overcome these problems and facilitate stronger econometric testing and more robust results with respect to structural factors that may be influencing wages growth. These data sources may also allow for future work in this area to be centred around an underlying theoretical framework that allows for heterogeneous productivity growth across firms and associated changes over time in industry concentration and profit mark-ups which generates a set of testable hypotheses. In the context of the central focus of this paper, these data sources allow work to be undertaken on whether one of the key findings with respect to new technology – namely, its very uneven take-up across firms – is an important part of the explanation for low average nominal wages growth.

A shift to a more micro-level framework that generates a set of testable hypotheses can be seen in the overseas work linking this uneven take-up across firms of new technology to the widely observed trend decline in labour's share of national income. These studies are examined in Section 3 below. Section 4 sets out some a priori reasons why some aspects of the theoretical framework used in these studies of factor income share movements may have relevance to explaining why nominal wages growth has been so low in many countries, and sets out some suggested areas for further research on this front.


Baldwin (2016) differentiates the ‘old’ globalisation, associated with invention of the steam engine and the major decline in the cost of transporting goods across countries, from the ‘new’ globalisation associated with the ICT revolution and the dramatic fall in the cost of transferring information across countries. Whereas old globalisation was focused on separation of production and consumption, new globalisation is focused on unbundling production processes via supply chains. [6]

See, for example, Timmer et al (2014) and Francois, Manchin and Tomberger (2015). For Australian studies, see Kelly and La Cava (2014) and Heath (2017). [7]

See, for example, Campbell et al (2007); Flecker (2010); Hübler and Hübler (2010); Degryse (2016); Goldschmidt and Schmeider (2017); Haldane (2017b); and Hudson-Sharp and Runge (2017). For some Australian studies, see Lass and Wooden (2017) and Mooi-Reci and Wooden (2017). For studies specifically on the impact of automation on wages growth and wage inequality, see references in Autor and Salomons (2018). [8]

This point is disputed by Gordon (2016, pp 444–448). [9]

This finding on key factors inhibiting take-up of new technology in the legal industry is consistent with much broader feedback across many sectors of the Australian economy from the RBA's business liaison program; see Lai, Poole and Rosewall (2018). [10]

Another factor examined in RBA research is the frequency, average size and dispersion of wage increases. Bishop and Cassidy (2017) found that all three had declined in recent years. However, this is more a description of what is going on behind the aggregate numbers rather than an explanation of what is driving surprisingly low nominal wages growth. [11]