Resolving the productivity paradox

https://doi.org/10.1016/j.matcom.2008.01.029Get rights and content

Abstract

Solow [R. Solow, We’d Better Watch out, New York Times Book Review, 1987, p. 36] made the statement that ‘we see computers everywhere except in the productivity statistics’. This has come to be known as the “productivity paradox”. Whether this is in fact a paradox or a direct implication of the diffusion of technical change is the focus of this paper. In particular, the implications of two different theoretical treatments of technology diffusion in an economy are considered; the traditional model of [R. Solow, A contribution to the theory of economic growth, Q. J. Econ., 70 (1956) 65–94] and the alternative view of [R.G. Lipsey, K.I. Carlaw, C.T. Bekar, Economic Transformations: General Purpose Technologies and Long Term Economic Growth, Oxford University Press, Oxford, 2005]. These two distinct views articulate two general empirically testable hypotheses that are captured in a number of specific tests including measures of the diffusion of information and communication technologies (ICT). Although weak, the evidence supports the view of [R.G. Lipsey, K.I. Carlaw, C.T. Bekar, Economic Transformations: General Purpose Technologies and Long Term Economic Growth, Oxford University Press, Oxford, 2005].

Introduction

This paper seeks to demonstrate that the apparent “productivity paradox” is a creation of the modelling approach taken to explain the role of technology in an economy. The paper focuses, in particular, on economic growth caused by information and communication technology (ICT) which is regarded as a modern general purpose technology (GPT). The following question is also raised “has ICT caused a revolution in global production and communication, or not?”. The answer to this question lies in separating the diffusion of this technology from measured output or productivity gains generated by it. There seems to be little disagreement that computers, the Internet and the myriad supporting complementary technologies that they have enabled, have revolutionized production taking the world into the age of the global economy. What is debated is whether this technological revolution is having the kinds revolutionary influences on economic growth that were witnessed with the First and Second Industrial Revolutions, themselves based on the technologies of automated textile manufacturing and steam in the case of the First and electricity, machine tools and chemicals manufacturing in the case of the Second. The view proposed here is that in order to become productively useful all technological knowledge must become embodied in some real physical component of the work whether it is physical or human capital (including all tacit skills), laws and legal institutions, or social and cultural norms. This is why we do not immediately see the benefits of new technologies in the National Accounts. Only when these new technologies have been sufficiently diffused to actually register in the accounts, do we actually ‘see computers everywhere’.

Furthermore, each of these embodiments requires costly investment, so the separation of the contribution of technological change from measured factors such as physical and human capital to economic growth is difficult. The key to connecting technological change to economic growth lies in identifying specific embodiments of new technology and determining their contribution to economic growth over a long horizon.

The debate about technologies’ contribution to economic growth is currently focussed on ICT's impact on economic growth. The so called productivity paradox is central to this debate. In essence it is a combination of a number of stylised and anecdotal observations about the proliferation of computers and ICT with the statistical observation of a decline in the growth rate of total- or multi-factor productivity (TFP or MFP) in many OECD countries, starting in the early 1970s and running through to the middle of the 1990s. The erroneous presumption that underwrites the paradox is that TFP measures technological change in a perfectly, contemporaneously correlated fashion. One view in this debate holds that the paradox has been resolved by the emergence of the New Economy in the United States as evidenced by the measured increase in TFP growth starting in the mid 1990s. An alternative view is that there is no paradox at all because the productivity statistics show that no technological revolution has occurred. We take these two views as being representative of what we call the traditional view of growth driven by technological change. This is a view that is typified by use of “the aggregate production” function first introduced by Solow [21] in which technology is captured by an exogenous shift parameter, is unstructured and has a contemporaneous, positive impact on output.

An alternative view is that there is no paradox because there is a real technology cycle that causes real productivity slowdowns. In line with this view a number of students of general purpose technologies (GPTs) argue that the introduction of new GPTs can cause large structural adjustment costs as the economy exploits the new technology see for example [1], [12], [13], [14], [17], [18]. These theoretical views reconcile the observed facts of large-scale technological change with initial declining productivity numbers by noting that some technological change brings with it a costly adjustment process. Lipsey et al. [18] argue that the pattern is not necessarily inherent in the new GPTs themselves, but it is a possible outcome of the interaction between new GPTs and the existing economic structure into which they are introduced. If there is sufficient friction between the new technologies and the existing economic structure, including necessary redesigns of physical capital, reskilling of human capital and changes in the organizational technology of firms then a real productivity slowdown can follow the introduction of a transforming GPT for a time. But the introduction of the GPT ultimately rejuvenates growth and there is a long term productivity benefit. We call this third view the non-traditional view.

The traditional view of growth and technological change has an immediate and easy to test hypothesis. Output growth and technological change are contemporaneously and positively correlated. This view expects to observe a positive correlation between the diffusion of a new technology and measured productivity growth rates so there is a paradox for those in the traditional view that observe the proliferation of ICT but no productivity boom until late in 1990s.

The alternative, non-traditional view generates the testable hypothesis that a new technology's impact on growth will not be immediately positive and that it can potentially can initially cause productivity slow downs which will be turned around as the technology mature. So we should expect to observe no correlation or even a negative correlation between technological diffusion rates and productivity growth rates.

In this paper we examine what if anything the New Zealand data tell us. Our data is limited causing our conclusions to be more conjecture then final statements. What we do see is some support for the non-traditional view in the New Zealand data.

Section snippets

New Zealand ICT diffusion and productivity

The contributions of embodied technological change to TFP growth have been studied in the growth accounting literature. Hulten [15] and Jorgenson [16] have focused on the measurement of the efficiency of the capital stock and the effects of measurement errors on productivity estimates. These authors argue that quality change (or Investment Specific Technological (IST) growth) is difficult to observe, and therefore may not be measured accurately in the National Income and Product Accounts

Conclusions

We set out in paper to analyse two views of technology diffusion in the context of ICT diffusion in New Zealand. In doing so, we begin the development of a theory of MFP or TFP by developing a multi-sector model of endogenous GPT-driven growth. The need for such a theory arises out of the mutually incompatible interpretations of technological change and productivity change. Such a need also arises out of the inconsistency in the interpretation of TFP growth as a measure of technological change

Acknowledgements

Research support from the FoRST-funded PGSF contract UOWX0306, Impacts of ICTs on Work and Communities is gratefully acknowledged.

References (22)

  • K.I. Carlaw et al.

    Externalities, technological complementarities and sustained economic growth

    Res. Policy

    (2002)
  • J. Greenwood et al.

    The role of investment-specific technological change in the business cycle

    Eur. Econ. Rev.

    (2000)
  • P. Aghion et al.

    On the macroeconomic effects of major technological change

  • K.I. Carlaw, ICT in Australian Economic Performance: An Investigation of the Assumptions Influencing Productivity...
  • K.I. Carlaw et al.

    The sources of total factor productivity growth: evidence from Canadian data

    Econ. Innov. New Technol.

    (2004)
  • K.I. Carlaw, R.G. Lipsey, GPT-driven, endogenous growth, Econ. J., in...
  • K.I. Carlaw, R.G. Lipsey, Externalities versus Technological Complementarities: a model of GPT-driven, sustained...
  • K.I. Carlaw et al.

    GPT-driven endogenous growth

    Econ J

    (2006)
  • P. David

    The dynamo and the computer: an historical perspective on the modern productivity paradox

    Am. Econ. Rev.

    (1990)
  • R.J. Gordon
    (1990)
  • J. Greenwood et al.

    Long run implications of investment-specific technological change

    Am. Econ. Rev.

    (1997)
  • Cited by (8)

    View all citing articles on Scopus
    View full text