5  Firm and Industry Knowledge Spillovers

The third mechanism whereby increases in an individual’s skill levels might give rise to improvements in the productivity of firms is through their direct effect on other workers or other firms. These complementary effects stem can give rise to knowledge spillovers. They will increase the returns to inputs in the firms they influence. A more complex interaction occurs at the national level where poor expectations may be influencing decisions about longer term investments in technologies and skills, and leading to lower productivity outcomes. These three avenues are considered in turn. (R&D spillovers are not discussed.)

5.1  Within firm effects

Knowledge complementarities within the firm can take a number of forms. The actions of entrepreneurs and senior managers, including their failures, can have effects across firms or down hierarchies. Workers are more productive where there is an existing stock of knowledge, either in the form of experienced and skilled staff, or of knowledge management systems. Mentoring facilitates knowledge transfers and supports learning by doing. Effective knowledge management routines reduce search costs and assist problem solving. Working with specialists in teams utilises the complementary skill bases held by team members. Some portion of these knowledge transfers or processes may be priced in terms of wages or reflected in the intangible value of the firm, and be internalised within the decision making of the firm. Knowledge spillovers are externality effects that directly improve the productivity of other workers and firms. They may be priced or unpriced. Their extent is hard to gauge.

A person’s earnings can be influenced by “who you are” (their individual characteristics), “where you work” (firm effects), and “who you work with” (co-labourer effects). A worker’s earnings progression does not depend only on the abilities they bring, but also the environment in which they work. Abowd and Kramarz (1999) report evidence on the size of firm effects. These range in some studies from being quite low, up to the point where the firm effect can explain as much of earnings variation as do personal characteristics. In the former case, the firms in the industry may be more homogeneous and so changing jobs will do little for earnings, or firms may be more segregated in the types of skills that they employ. Higher firm effects suggests that significant productivity differences occur between firms, perhaps because of organisational structure or management practices. Lengermann (2002) explores the co-labourer effect, where the presence of other skills in a team raise the worker’s productivity. His analysis sought to estimate the influence of the average characteristics of other workers in a firm on an individual employee’s wages. Further, as Blundell et al (1999b, p6) report, individuals working in an industry experiencing technological progress can also receive higher returns to education. In this literature it is possible that the measure of worker characteristics is not picking up the full range of abilities that are valued by firms, with the other effects being over-estimated.

The processes and practices that managers put in place can also substantially influence workers’ effectiveness. The management models described in S 3.2.1 note a range of effects. Ichniowski and Shaw (2003) refer to the differences in the “connective capital” of workers brought about by differences between innovative and traditional human resource systems. In the former, a worker’s access to the knowledge and skills of co-workers is important in teamwork and problem solving. Similarly, operating routines and dynamic capabilities are used to synthesise information into effective internal technical, organisational and managerial processes that enable a firm to maintaining competitive advantage in a changing environment. Thus, effective managers increase the productivity of their subordinates.

Battu, Belfield, and Sloane (2001) look at knowledge spillovers within the workplace. Using the UK Workplace Employee Relations Survey and data on workforce characteristics of firms, they provide evidence on how own earnings are affected by co-workers’ education. They find that there is a positive and substantial association between own earnings and mean workplace years of education, but negative interactions between own and co-worker education levels. They suggest that education is less valued at workplaces where education levels are already high. A question with this study is how effectively other firm-specific effects, like organisation or management practices, are controlled for.

5.2  Knowledge spillovers between firms

Knowledge spillovers can also occur between firms in an industry (the so-called Marshal-Arrow-Romer (MAR) knowledge externalities).[13] Where there is a greater the stock of shared knowledge in an industry, the individual or firm will be more productive. This can occur generally across an industry through the diffusion of R&D knowledge embedded in better technologies and practices. Externalities can be associated with experimenting with organisational design, as successful designs are relatively easy to imitate. Locally, there can be a geographic concentration of specialisation (at the firm level, or in the workforces employed) which improves the matching of inputs in production, and the achievement of economies of scale.

Duranton and Puga (2003) identify three main mechanisms giving rise to agglomeration effects. They are the sharing of inputs (where indivisible facilities can be supported by the geographical industry concentration), matching (through improving frequency and quality of transactions in input and output markets), and learning (where proximity increases the level and quality of knowledge generated and transferred between firms). A large indivisible facility provides increasing returns to scale, and lower costs of services to its users. Thicker labour markets with more agents and transactions increase competition, lower search costs, and allow better matches between firms and the skills required. Better matching arising from “thick” input markets generates externalities, in that it allows increased firm specialisation with economies of scale in production, and permits specialisation by workers. Thicker markets also lessen bilateral relationships between buyers and sellers, and the resulting risks of hold-up. Geographical proximity encourages knowledge spillovers by lowering transaction and communication costs. Increased face to face contacts are helpful for the transfer of tacit information. Proximity to other skilled individuals assists knowledge acquisition, thereby aiding knowledge diffusion through an industry. The greater inter-dependence and exchange amongst firms in a localised industry also appears likely to enhance absorbtive capacity and stimulate innovative activity.

Crawford (2004) summarises the econometric evidence on the size of agglomeration effects. Several studies are cited that identify economically significant effects from agglomeration on labour productivity. These effects seem to operate primarily through linkages between denser urban areas and human capital. Evidence on the relative importance of the three specific mechanisms giving rise to these effects is more difficult to establish. Dumais, Ellison, and Glaeser (1997) found evidence that labour market specialisation was by far the most significant mechanism. Using data on firm dynamics in US manufacturing industries, they found that this effect appears to arise from the importance of shared labour markets, where the occupational mix of the firm is closer to that of its potential customers and suppliers. It has proved difficult to distinguish the knowledge spillovers operating within and between industries. However, Audretsch and Feldman (2003), in exploring the spatial dimension of knowledge spillovers and innovative activity, report evidence showing geographical concentrations of patent citations and product innovations, after controlling for industry agglomeration effects. Co-agglomeration effects between industries vary greatly, with only three resourced based industries showing significant effects in terms of accessing inputs from a single supplier.