5.3  Path dependence in skill/technology choices

A related line of inquiry concerns the possibility, at the industry or national level, of low skill levels and skill shortages leading to a low skill/low technology equilibrium. A number of influences may be present. Several of the firm dynamics findings noted earlier point to potentially significant path dependence in skill/technology choices at the firm level. The rate of diffusion of ICT will be influenced at the firm level by the costs of adoption, which will include what they can learn from other firm’s experience of implementing the technology. Persistent skill shortages may also have implications for capital deepening. With sunk investments in capital, labour, technology and organisation, and significant adjustment costs, it may be costly and demanding for a firm to change its input mix while the benefits may only appear over the long haul. The challenge is to understand better what would generate and sustain a low skill/low technology equilibrium, and whether these effects are significant in practice for policy purposes.

Acemoglu (1996) and Redding (1996) provide models of the choices of skill level and technologies that operate nationally. In Acemoglu’s model, firms and workers make investments in physical and human capital before production begins. Matching of workers and firms is assumed to occur randomly, possibly because search is costly. Their investments are made before the match is known. The model yields the result that an increase in the average level of human capital can yield increasing private returns. This is because the expectation by firms of higher skill levels encourages them to make greater capital investments. With random matching, some workers with lower skills will be matched with firms making higher capital investments. This yields an external effect. Paul David (2001, pp37,38) notes two limitations of Acemoglu’s analysis, both related to the random matching assumption. First, there may be ways of increasing the co-ordination of expectations, such as providing market information, or increasing the level of work-based training. Second, the mismatching of workers and firms, and consequent wasted investment, is likely to yield risk averse behaviour to limit it, such as devising selection rules.

The model of Redding (1996) investigates the effects of joint decision making by workers about investment in education, and by entrepreneurs in terms of their investments in R&D. In Redding’s model, workers are again randomly matched with entrepreneurs. This inter-dependence again allows the possibility of multiple equilibria, with the outcome depending on expectations about the other party’s level of investment. One outcome is low levels of skills and innovation, which reflects a failure of coordination. Alternatively, credible commitments to raising skill levels could also raise R&D expenditure. This suggests that skill acquisition and R&D are strategic complements. Governments may, therefore, have a role in coordinating expectations. One testing of Redding’s model is provided by Nickell and Nicolitas (1997). They investigate whether there is any connection between the shortage of human capital and rates of capital investment or R&D. They use data for UK manufacturing firms for the period 1976-94, and on the proportion of firms whose output is constrained by shortages of skilled and unskilled labour. Their finding is that a permanent 10 percentage point increase in the number of firms reporting skilled labour shortages led to a permanent 10% reduction in capital investment and a temporary 4% reduction in R&D expenditure. A more recent UK study using firm level data by Forth and Mason (2004) finds that firm performance is improved by the rapid adoption and use of ICT, but this is hindered by deficiencies in skills for adopting ICT.

Other evidence that incentives to invest in human capital and technology are interdependent comes from a number of comparisons of UK, US and European industry performance, in particular the studies undertaken by the UK’s National Institute for Economic and Social Research (NIESR) (and reported briefly in S 3.3). These studies document and seek to explain the wide differences in the technological strategies being followed by countries in similar industries. Following Finegold and Soskice (1988), a number of authors have argued that the UK is trapped in a low skill equilibrium. Keep and Mayhew (1999) and Caroli (2001) argue that the availability of a skilled workforce is a significant factor in the persistence of a low skill/technology strategy. The suggestion is that it is a systems failure that requires a coordinated response across a range of policies and institutions. The dominant mechanism proposed is the kind referred to by Acemoglu where there is little demand for skilled workers, the incentives to train are reduced, while the limited supply of skills influenced firms’ technology investment decisions. Some (such as Caroli (2001)) put more stress on poaching externalities, arguing that the adoption of new technologies requires a higher proportion of general skills, which firms have less incentive to invest in, and which must therefore be provided by the general education system. Still others utilise an industry spillover story, where individual firms wanting to break out of a low skill equilibrium would need to finance a disproportionate share of R&D, as beneficial externalities from other firm’s R&D activities are absent.

One might provisionally conclude that there is a case to be answered as to why different technological trajectories are sustained over time, for example between the UK and Germany. However, more rigour appears needed in identifying the key factors hindering the transition out of a low skill equilibrium. Little comment is made on the differences in labour market regulations in the UK-Europe comparisons, for example, and whether wage compression in Europe is making a significant difference. Nor is it clear why increasing the supply of skilled workers would be a major policy option for addressing the issue.

What gives rise to low technology/skill matches is clearly important. The question is how these technology diffusion and adjustment processes would benefit from governments taking a greater role in coordination. The evidence is less persuasive that a systems failure is involved that requires a coordinated response across a range of policies and institutions. There is evidence for the UK that persistent shortages of skilled labour may be influencing technology adoption and capital investments. The case for further increases the supply of skilled workers depends on whether other factors are more critical. More analysis is needed of the rationale and evidence for path dependence, and of how governments might make an improvement. There may, for instance, be significant costs associated with firms changing their input mix, given sunk investments in capital, labour, technology and organisation, leading to slow rates of adjustment. The policy specifics here will be influenced by the economy being analysed.