Knowledge Accumulation and Diffusion: Analysing Heterogeneity in an Interconnected World

Innovation is undoubtedly an important driver of long-run economic growth. Relatively recent innovations such as the PC or the Toyota manufacturing system offer clear evidence for a close link between innovation effort in the form of research and development (R&D) investment and successful development. The inability of an innovating firm to appropriate all of the knowledge it created and the 'spillover' of such knowledge to other firms leads to the well-known market failure of underinvestment in innovation and represents a prime justification for policy intervention. But should some industrial sectors be deemed more important than others not only because they innovate a lot (creating profits and jobs), but precisely because the knowledge they create is believed to spill over to other parts of the economy? How can we measure these unobservable spillovers, how important are they, and to what degree do they extend across national borders? Does the relative importance of specific sectors warrant subsidies, incentives and protection?

The notion of 'industrial policy,' the selective government support for specific sectors, is presently being widely debated again among OECD country think tanks and policymakers as well as within international institutions such as the World Bank and UNIDO. A renewed interest among policymakers in the patterns of 'knowledge accumulation' and spillovers, and the means to benefit from domestic and international knowledge investments for economies at different stages of development is the prime motivation for this research.

The investigation of returns to knowledge investment and spillovers has exercised researchers since Griliches' seminal 1964 paper added R&D stock to a productivity analysis of US agriculture. Building on previous work by the PI the projects contained in this proposal represent a significant departure from the conventional empirical approach and emphasize:
(1) that the analysis of knowledge spillovers deals with unobserved phenomena. Analysis needs to demonstrate that estimated spillover effects do not capture global economic shocks, spillovers unrelated to knowledge diffusion or other heterogeneities perhaps resulting from the choice of empirical model. The existing literature typically imposes one of two 'transmission channels' for knowledge to spill from the innovating country or sector to the rest of world/economy: (i) a trade and FDI channel, or (ii) a channel related to geographical distance. This is highly restrictive, does not rule out contamination by the above effects and commonly provides no test for the relative importance of these channels. The projects adopt more flexible panel time series models and place the explicit testing of rival hypotheses at the centre of the empirical approach.
(2) that economies at different stages of development may have different 'needs and tastes' for knowledge. The assumed potential for spillovers from high-tech innovators in rich countries to agrarian societies in poor countries runs counter to any notion of 'appropriate technology,' a disconnect which due to agro-climatic differences is particularly salient in agricultural innovation. The projects recognise this potential heterogeneity, both in terms of differing patterns across industrial sectors as well as within sectors across countries. Providing more detailed results represents a significant step away from reporting average effects across the entire sample of countries or sectors.

Policymakers in developing and developed economies alike are keen to gain insights into the magnitudes and patterns of knowledge spillovers and to learn about effective strategies to appropriate relevant knowledge for sustainable development. The findings from this research provide important insights into the macroeconomic performance of developing and developed economies, and highlight the potential for policy intervention for poverty reduction and the safeguarding of competitive advantage.

The questions addressed in this proposal are currently widely debated in policy circles in terms of (a) the re-emergence of industrial policy in OECD countries; (b) the recognition within institutions such as the World Bank and UNIDO that the industrial structure (agriculture, manufacturing, services) of less developed economies and the production and transfer of knowledge play central roles in the development process; (c) the concern over an increasing misalignment in agricultural R&D between developed and developing countries voiced at the G20 summit in Cannes, meaning that significant knowledge spillovers from the former aiding productivity increases in the latter may no longer materialise.

In the UK the insights and conclusions from this research will be of direct interest to researchers and practitioners in the public policy sector, including the Department for Business, Innovation and Skills, the National Institute of Economic and Social Research, the UK Intellectual Property Office, nesta, the Department for International Development, the Overseas Development Institute as well as NGOs, think tanks and interest groups in the fields of innovation and international development.

Beyond the UK, findings speak to researchers and practitioners in the public administration of developed and developing economies as well as international organisations such as the OECD, UNIDO, the UN Food and Agriculture Organisation (FAO), and the Consultative Group on International Agricultural Research (CGIAR).

Patterns and magnitudes of knowledge spillovers are of great interest to policymakers in developed and developing economies. However, leaving aside concerns over the identification of true spillover effects, the insights gained from existing research are typically limited to relatively broad statements, which frequently do not lead to policy-relevant recommendations, e.g. "on average the combined impact of R&D investments in six [OECD] countries close to the world's technology frontier [on other OECD countries]... is at least three times as large as that of [the latter's] domestic R&D" (Acharya & Keller, 2009). It is not clear that 'stop domestic investment in R&D' represents a sensible strategy for sustainable development. 'Transmission channels' through which knowledge was said to diffuse were typically not so much explored as imposed, leaving limited lee-way beyond a stark 'yes' or 'no' for judging the importance of alternative channels.

The empirical innovations applied in this research put considerable emphasis on working out both commonalities and differences across countries and sectors:
- in the OECD project research will establish (i) the interconnectedness between countries and sectors in terms of economic shocks; (ii) the major knowledge sources (production and diffusion) amongst countries and sectors, (iii) the relative importance of various transmission channels for different countries and sectors, recognising they may employ different technology;
- the LDC manufacturing project will provide insights into (i) what forms of (non-R&D-measured) innovation have proved productivity-enhancing for different countries, (ii) what role, if any, (high-tech or other) knowledge creation in OECD countries can play in LDC manufacturing development, (iii) the relative importance of various knowledge transmission channels for economies at different stages of development;
- the LDC agriculture project will establish (i) the private returns on investment in agricultural R&D taking account of global shocks and spillovers, (ii) what role, if any, agricultural R&D in OECD countries plays for agricultural productivity in LDCs, as well as (iii) changes in these relationships over time.
There is thus considerable potential for the proposed research to address the heterogeneous needs for insights into the patterns and magnitudes of knowledge accumulation and spillovers in diverse sets of countries as developed above.