In this preface, I first ask: What distinguishes this book from the many others on technology and on economic growth? I then go on to consider the evolution of the three most important themes of this book.
Why Another Book on Growth?
While books on long-term growth have always been popular, the last few years have seen a great increase in their number. Readers may wonder if ours is simply another marginal addition to this fast-growing literature. We think not. First, we are interested in the phenomenon of general purpose technologies (GPTs) in themselves and we say much more about them, both descriptively and analytically, than is typical in books on growth. Second, in our work on growth we take up a broader set of themes and employ a larger array of analytical tools than is typical of most books on growth and technological change. While others concentrate on one or another of these techniques, we do not hesitate to use historical analysis, formal modeling, simulation techniques, what Richard Nelson calls appreciative theorizing, and aspects of evolutionary economics.
In Chapter 1, we outline our coverage in detail and note much of what we have to say that is new on each of these topics. Here we merely illustrate the wide scope of our coverage, which, for better or for worse, is one of the distinguishing features of our book. We start by observing that long-term growth is driven mainly by technological change. This leads us to study the nature of technology and how it changes, building on material found in books such as Rosenberg's "Inside the Black Box." We argue that understanding technological change requires an evolutionary approach, such as was pioneered by Nelson and Winter in "An Evolutionary Theory of Economic Change." We outline such an approach and contrast it with neoclassical theory. Because over the centuries new technologies radically alter more or less everything in the socio-economic order, doing much more than just increasing output per person, standard neoclassical theory is a relatively poor tool for studying their effects. We argue that one approach that handles these effects well is a combination of institutional and evolutionary economics that we call structuralist-evolutionary (S-E) theory. The contrast between neoclassical and S-E theory leads us to consider two different world views of how the economy works and of what policies are effective in achieving given ends.
We also argue that a full study of growth requires an understanding of quite a bit of the history of technological change as is found in Mokyr's Lever of Riches. Here we concentrate on the big shocks caused by GPTs as are discussed in Dudley's The Word and the Sword (although he does not use the term "general purpose
Technologies"). Big GPT shocks change almost everything in a society and revitalize the growth process by creating an agenda for the creation of new products, new processes, and new organizational forms. To elaborate, we study GPTs through an S-E lens, spending much time developing an S-E theoretical structure in which we situate GPTs. We systematize much more of the knowledge of how GPTs evolve and affect the society than we were able to do in the two chapters that we contributed to Helpman (1998).
We then discuss the nineteenth-century emergence of sustained growth of output in the West, building on the analyses in Rosenberg and Birdzell's How the West Grew Rich and in Landes' The Unbound Prometheus, but putting much more emphasis on science than is usual. This leads us to ask why sustained growth of output was not generated endogenously outside of the West, where we use much of the analysis found in Toby Huff's The Rise of Early Modern Science, and take issue with some of the arguments in Kenneth Pomeranz's The Great Divergence. Then we turn to the emergence of the West's sustained per capita growth that happened later in the nineteenth century. This leads to a discussion of population dynamics as is found in Easterlin's Growth Triumphant, although, in contrast to his appreciative theorizing, we build our analysis around several simulation models that make use of neoclassical growth theory.
We argue that once sustained growth has been established, we can learn quite a bit about its dynamics from formal models of GPT-driven growth. We develop new ways of theorizing formally about GPTs that allow us to incorporate much more of their richness than was possible in the first-generation models, which were based on crude assumptions needed for theoretical tractability. In doing this, we are taking up the program that we enunciated at the end of our contribution to the Helpman volume and that we thought would by now have been much further advanced than it is.
"Developing satisfactory theories of GPTs is not a task that will be completed quickly or easily. It seems to us that the theoretical research program should be to extend existing models, and/ or to develop new models, to capture more of what we know empirically about GPTs rather than elaborating and generalizing just because we are able to do so. In this program there would be a large payoff to the development of new models that are designed to capture more of the characteristics of GPTs in their assumptions, and then explore the implications of those assumptions." (Lipsey, Bekar, and Carlaw 1998: 217-18)
Finally, we discuss some policy implications of our approach to understanding long-term growth.
Doing all of what we have just outlined requires that we cover a much wider range of topics, using a larger variety of tools, than is found in almost all other books dealing with growth and/or technological change. We hope that we have at least begun the process of integrating these various topics and tools into a coherent analysis of both the causes and the consequences of long-term growth.
Evolution of Three Important Themes
Three of the book's most important themes were a long time in gestation: (a) the relation between long-term economic growth and general purpose technologies (GPTs); (b) the importance of science in the First Industrial Revolution that initiated sustained long-term growth in the West; and (c) the relation between the twentieth-century revolution in information and communication technologies (ICTs) on the one hand and the so-called productivity paradox, and the use of total factor productivity (TFP) to measure technological change, on the other hand.
GPTs and Long-Term Growth
Historically, this book began with my investigation into the causes and consequences of long-term growth. I quickly discovered Perez and Freeman's concept of a techno-economic paradigm, which my co-authors and I came to see as a key to understanding the impact of technological change in terms of periodic transformations of the economy. After using the concept for a few years, we developed our own more focused concepts of transforming technologies and the facilitating structure. Later, we discovered that our concept of a transforming technology was more or less the same as that of a GPT that had recently been put forward by Bresnahan and Trajtenberg, so we switched to using that term. Our work then developed into two distinct but interrelated research programs concerning (a) GPTs and (b) long-term economic growth.
We sought to understand what GPTs were, how they evolved, and how they impacted on the economy. Since the concept of a GPT was introduced into the literature just over ten years ago, it has received a growing amount of attention with many scholars utilizing it in their research. Unfortunately, on the theoretical side there have been no further advances, either in modeling it or in delineating its extent empirically, since Helpman's 1998 volume (in which the present authors have two chapters). We believe that modeling has not been expanded beyond the crude first-generation models found in Helpman because the standard theoretical maximizing techniques applied to GPTs quickly become intractable when elaborations are made in the direction of increasing realism. We sought methods of breaking through the roadblock that was so created. What we regard as a success in this endeavor came when we developed simulation models of GPTs which, although much less elegant than analytical models, are not constrained in the same way and can handle any degree of complexity that is needed to incorporate into formal models a large set of typical, GPT characteristics. Some of the many pay-offs to this approach are developed, in Chapters 14 and 15.
Unfortunately, on the empirical side there is considerable misunderstanding on just which technologies are and are not GPTs. For example, Moser and Nicholas (2004) argue that electricity, one of the most pervasive GPTs of all time, is not a GPT at all. The questions of what a GPT is and how to identify one are taken up in detail in Chapter 4.
On long-term economic growth, we sought to integrate the concept of GPTs into the historical story of growth and to use our new insights to investigate how the episodic growth that had existed for millennia was transformed in the nineteenth century into sustained growth. In writing the book around the theme of long-term growth, we may have obscured the contributions we seek to make through our research program to better understand and model GPTs. We hope that this is not so since much of what we say about GPTs, particularly in chapters 4, 5, and 6, can be divorced from considerations of very long-term growth.
Science and the Industrial Revolution
The second main theme is the importance of mechanistic science in the First Industrial Revolution. No one doubts that science was important in the Second Industrial Revolution and that it grew more important as a driver of invention and innovation as the twentieth century progressed. But the prevailing view in the 1990s seemed to be that, up until the late nineteenth century, empirically based technological advances led science (by, among other things, presenting scientists with such problems as understanding fermentation and heat transference), not the reverse.
I first began thinking about trajectories in the advance of scientific knowledge when 1 encountered Joseph Needham's argument that, although the Chinese did not have Newtonian mechanics, their holistic approach might have allowed them to jump directly to twentieth-century quantum mechanics. It was clear to me that although he was a great scholar of Chinese technology, Needham could only have held that view if he knew very little about how scientific knowledge grows cumulatively. This led me to think about the place of Western science in the emergence of sustained economic growth at the time of the First Industrial Revolution. Nathan Rosenberg had argued that on balance, up until late in the nineteenth century, technology led science, not vice versa. While arguing with him when I presented some of my preliminary thoughts to the CIAR group (discussed in the Foreword), I had a great insight. He, and many others who argued in a similar vein, were thinking of modern science: great embracing hypotheses from which specific applications were deduced. But this was not the nature of early modern science. At that time, the overarching hypotheses were due to Aristotle, whose science had been fully integrated into Christian theology by the great scholastic philosophers of the Middle Ages. Early modern science can then be seen as a piecemeal testing and gradual refutation of Aristotelian science. Not until Descartes and Newton was Aristotle replaced by new overarching scientific world views.
With these concerns in mind, I reread Mokyr's "The Lever of Riches" and found him dismissive of the importance of science at the pre-industrial stages of technological history. But what caught my attention were his statements that "Britain did not have a significant scientific advantage that would explain its technological leadership" and that "Britain had no more science than the Continent, only different science" (Mokyr 1990: 242). That was the clue: it really did matter that only Britain had Newtonian science, while France had Cartesian science, and those outside of the West had neither. Britain had a significant advantage in Newtonian mechanical science, which was what mattered for the First Industrial Revolution, and the great eighteenth-century engineering works that preceded it. Two books were critical in my elaboration of this view: Toby Huff's "Rise of Early Modern Science," and Edward Grant's "The Foundations of Modern Science in the Middle Ages." Later, Margaret Jacob's "Scientific Culture and the Making of the Industrial West" filled in a missing piece of the puzzle by showing how much Newtonian science permeated the whole of eighteenth-century British thinking. A detailed study of developments in science and technology in the early modern period filled in the remaining blanks.
So, understood as referring to modern science, the statement that technology presented results to be used by science rather than vice versa is correct. (Rosenberg 1982: ch. 7 gives half a dozen examples.) But understood as referring to science as it was in the early modern period, the statement is questionable. This led to a detailed study of the mutual interaction of early modern science and technology. I presented these ideas to our CIAR group in 1997.
Then in 1999, Clifford Bekar and Kenneth Carlaw joined me in preparing a paper on this issue for a conference entitled "On the Origins of the Modern World: Competitive Perspectives from the Edge of the Millennium" in Davis, California. (1) Stated in a nutshell, our theses were: (a) early modern science and technology coevolved without one being the clear leader of the other; (b) Newtonian mechanics, the first fully modern, overarching, scientific `laws' were critical to the First Industrial Revolution, which helped to explain why it occurred where and when it did (in eighteenth-century Britain); (c) the absence of Newtonian science solved the puzzle of why China, which was the equal of Europe in so many other ways, failed to generate its own indigenous industrial revolution.
We were roundly attacked by the assembled ranks of Sinologists, who accused us of being hopelessly Eurocentric, but equally encouraged by a group of technology students, who remained silent in open discussion but supported us privately. We wrote these ideas up and had them rejected by three major journals. Nonetheless, we were sure we were onto something because the referees' reports were divided almost equally between those who said the ideas were so commonplace that they should not be published and those who said they were so obviously wrong that they should not be published. Our answers to the Sinologists and our analysis of the importance of science to the Industrial Revolution are mentioned briefly towards the end of Chapter 1 and detailed in Chapters 7 and 8. We were also encouraged by some recent writings in which several authors have increased the importance they accord to science in the First Industrial Revolution. (2)
The Revolution in ICTs, the productivity paradox, and TFP
The third theme is actually a set of interrelated issues concerning ICTs, the productivity paradox, and TFP. Early on, I came to the conclusion that the world was experiencing a profound economic, social, and political transformation driven by the revolution centred around the electronic computer. When I first began to express this view in Canadian policy circles in the early 1990s, most economists were dismissive. Typical arguments against my view were:
"Technology changes more or less continuously, a little faster sometimes, a little slower at other times, but such change is not interrupted by the kinds of revolutionary events you describe. For conclusive proof look at the growth in total factor productivity, which measures techno-logical change, and which, if anything, has been slowing over the 1980s and early 1990s just when you assert the revolution was occurring."
This conflicting view set in motion three research subprojects within my general study of long-term economic growth. First, Bekar and I set out to study past technological shocks brought about by what we came to call GPTs. We identified twenty or so of these in all of history. (See Lipsey and Bekar 1995 for our first statement of these.) So we had established, contrary to my critics, that such transforming shocks have occurred in the past. This left us with our second project, to study the question: Is the so-called ICT revolution one of these or is it a lesser shock? Studies of its effects first published in Lipsey and Bekar (1995) left us in no doubt that it ranked with the most important of history's transforming GPTs. This led to our third project: to discover what was wrong with the commonly repeated argument that the deceleration in the rate of growth of the Solow residual, now called TFP, indicated a slowdown in technological change rather than a new technological revolution. This set Carlaw and me off on a search that extended over more than half a decade into the meaning and behavior of TFP As our understanding of these issues evolved, we presented them in a number of workshops and finally in a massive paper presented at the conference in honor of Nelson and Winter in Aalborg, Denmark (Carlaw and Lipsey 2001). We knew we were onto something when the paper presenting our analysis of TFP was enthusiastically endorsed by students of measurement such as Erwin Diewert and Alice Nakamura, although rejected by a leading journal. The paper has since been published in The Canadian Journal of Economics (Lipsey and Carlaw 2004). Our main conclusion in this paper is that since TFP does not measure technological change, there is no paradox in observing high rates of technological change and low rates of TFP growth.
More generally, however, we argue in this book that the whole expectation of an acceleration in productivity growth associated with a new GPT-and the assumption of a paradox when we see the latter but not the former-is a case of the `Emperor's New Clothes'. The argument concerning why we should not necessarily expect a new GPT to be accompanied by a productivity bonus, as well as an enumeration of changes that make the ICT revolution rank as one of the most important transforming technologies of all time, can be found in Lipsey (2002 b). It is substantially repeated here in Chapter 4 under the subheading "The Myth of the Productivity Paradox." - Richard G. Lipsey
(1) The paper was listed as "Science, Institutions, and the Industrial Revolution" by Richard G. Lipsey, Clifford Bekar, and Kenneth Carlaw.
(2) Since this preface was written, we have heard that our paper has been accepted by a journal (see Bekar and Lipsey 2005, forthcoming).
Rights: By permission of Oxford University Press. www.oup.com
|
