Innovation Policy and Canada's Competitiveness

by

Kristian Palda

The Fraser Institute, Vancouver, British Columbia, Canada



Copyright (c) 1991 by The Fraser Institute. All rights reserved. No part of this book may be reproduced in any manner whatsoever without written permission except in the case of brief quotations embodied in critical articles and reviews.

The author of this book has worked independently and opinions expressed by him, therefore, are his own, and do not necessarily reflect the opinions of the members or the trustees of The Fraser Institute.


Preface

IN 1984 THE FRASER INSTITUTE PUBLISHED my Industrial Innovation: Its Place in the Public Policy Agenda. When I first started on the present volume, I believed it to become a second, updated edition of Industrial Innovation. Yet it turned out to be so different a book, using perhaps only a fifth of the previous material, that it deserved a different title.

I would like to thank Michael Walker and Fred Mannix for their kind and effective support. The monograph benefited from my research association with Petr Hanel of Sherbrooke University and with my two Queen's colleagues, Bohumír Pazderka and Lewis Johnson. Another Queen's colleague, Klaus Stegemann, provided valuable information. Linda Freeman typed the manuscript quite admirably.

The most substantial help came from my son Filip, to whom I dedicate this book. Kristian Palda


About the author

KRISTIAN PALDA IS PROFESSOR of business at Queen's University, Kingston, Ontario. Born in Prague, Czechoslovakia in 1928, he completed his undergraduate education at Queen's University in 1956 and obtained his M.B.A. and Ph.D. from the Graduate School of Business of the University of Chicago, where he was the recipient of the 1963 Ford Foundation doctoral dissertation prize.

Professor Palda taught business and economics at the Ecole des Hautes Etudes Commerciales in Montreal from 1958 to 1962, when he was appointed assistant and then associate professor at the State University of New York at Buffalo. In 1965 he went to Claremont Graduate School, Claremont, California where he was professor of business economics until 1970. He was appointed to his present position at Queen's University in 1970. He held visiting appointments and has lectured widely in North America and Europe, especially in French-language areas.

Professor Palda's research interests, always coloured by economic analysis, have been devoted to two fields: the examination of advertising effects in commercial and political markets, and issues in technological innovation. He has published seven books touching on these areas, the latest two being Industrial Innovation and The Role of Advertising Agencies in Canada's Service Sector with the Fraser Institute. His most recent articles on campaign spending and technological innovation appeared in Journal des économistes and Prometheus, respectively.

In 1987 Professor Palda won a Queen's University prize for excellence in research. In 1991-92 he spent his sabbatical year at the Prague School of Economics.


Introduction

THE FIRST PURPOSE OF THIS VOLUME is to provide adequate information and guidance to understand innovativeness and the policies designed to stimulate it. Ample statistics and references to recent literature are offered so that the reader will be able to reach an independent judgement.

Technological innovativeness is conceived of as both the creation of new products and processes and the receptivity of enterprises to the adoption of new or improved equipment or methods. Just as industrial research and development is often a necessary activity in the creation or reception of new technology, so innovativeness itself is frequently a required ingredient in the achievement of competitiveness.

Yet to this day, most of the governmental industrial policy thrust in Canada, federal and provincial, is aimed primarily at R&D support and to some extent at aid to diffusion. The second goal of this volume is therefore to make it clear that innovativeness is not an activity that operates in a vacuum. Just as any other economic endeavour, innovativeness-and the necessity of staying competitive-demands managerial direction and investment funds as backing.

Yet managerial effort and investment capital will not be forthcoming if what we call "business conditions" are not right. An economy saddled with excessive anti-market rhetoric, regulation, taxation, and subject to wide swings in currency exchange and interest rates will simply not appeal to dynamic managers and venturesome investors. To believe that another tax concession or specific subsidy to this or that R&D project, that yet another pinpoint industrial strategy will open up the horn of plenty is naive. If not naive, then self-serving and rent-seeking. Canada has had for years and still has the most generous R&D-plus-innovation-diffusion tax support system of all leading industrial countries and yet has not progressed an iota in its overall research intensity.

For this reason, and for many others discussed in this book, we are strongly opposed to any increase in provincial and federal taxpayer support of industrial research and technology diffusion. Although this book makes an effort to present the subsidization viewpoints, we do not find them persuasive.

So while innovativeness is an essential ingredient of a competitive economy, the basic conditions for its further flowering in Canada are not present. It would take a fundamental reversal in governmental fiscal and infrastructure policies, especially on the provincial level, to offer the incentives to innovation which industrial policy is unable to provide.

Kristian Palda


Chapter 1 Innovativeness: The Principal Issues

The government's optimism about technology knows neither programmatic, partisan, nor ideological Linda Cohen and Roger G. Noll, The Technology Pork Barrel, Washington: The Brookings Institute, 1991, p. 1.bounds.

Introduction

TWO MAIN THEMES RUN THROUGH THIS MONOGRAPH. The first is technological innovativeness-its determinants and its repercussions. The second is government policies that sometimes facilitate, but more often impede innovation's genesis and diffusion.

Technological innovation finds its embodiment in new or improved products or processes. To earn the "innovation" designation such products or processes must undergo the test of the market, whether they are successful or not. Without the commercial try-out, we can only speak of "inventions" or improvements. The point of technological innovation is that it either widens the scope of customer choice (new products) or lowers the purchase price (new processes), or Techological as opposed to organizational or managerial innovation. The latter may not rely at all on changed product or process configurations. Corporate decentralization in the '20s is an example, though clearly it could not have happened without the automobile and the telephone to facilitate it.both. Thus, it enhances the economic well-being of the nation.

But these effects are not easily measurable on the level at which they count most, that of the final consumer. Thus, we tend to concentrate on measuring changes in cost and productivity both among innovating firms and their commercial customers, changes consequent upon innovation's creation and A moment's reflection shows that the consumer household is a small factory in which market purchases are combined with the services of domestic appliances and the household's time to produce final, utility-yielding goods. We can therefore equally speak of household productivity, substitution of inputs, etc. and imagine what, for instance, time-saving innovative equipment contributes to consumer welfare. But household activities are not statistically assessed since they are not a part of the monetized economy. We thus have a less clear picture of innovativeness at this level.

It is clear that the efficiency (or productivity) of commercial customers can be enhanced by the installation of equipment, such as machinery or computers, purchased from innovative suppliers. We call such a purchase an "innovation adoption," which is considered as vital to economic progress as "innovation creation." A firm's, an industry's, an economy's innovativeness can thus be envisaged as consisting of both types of activities-the creation as well as the adoption/diffusion of innovation. Both are equally important, but their relative shares will depend on a host of influences, such as the resource base of the economy or access to large markets.

Importance of Innovativeness

A first impression of the estimates of economic progress ascribed to technological innovativeness is shown in a table compiled from the Nobel prize acceptance speech of the American economist Robert Robert Solow, "Growth Theory and After," American Economic Review (1988), v. 78, p. 314. E. Denison, Trends in American Economic Growth, 1929-1982, Washington: The Brookings Institution, 1985.Solow. Table 1, derived from Denison's estimates-which are based upon Solow's pioneering methods in detecting determinants of economic growth-attributes one-third of U.S. growth in the private sector output between 1929 and 1982 to technological progress. This is the kind of figure which makes politicians salivate when they contemplate the usually dismal economic statistics detectable in their wake. If this sort of golden-egg laying goose could be reared by relatively inexpensive industrial policies, then budget-tightening measures could be discarded in favour of taxpayer-financed projects dear to every pressure group in sight.

Click here to view Table 1: U.S.A. 1929-1982

There are, of course, other ways of measuring the outcome of innovative activity. An example is an appraisal of the contribution of biomedical research to health production. Vehorn et al. attempted to measure the degree to which this type of research lowered mortality in the United States over the years 1930 to 1978. One of their regressions is shown in Box 1. It appears that, during that period, a 1 percent increase in research effort (measured by biomedical PhDs awarded 10 years previously) resulted on average in a 0.05 percent reduction in mortality rates. Research effort in this estimate accounted for 23 percent of the total of 91 percent explained in the movement of mortality rates.

Click here to view Box 1: Effects of Biomedical Research on Mortality, U.S.A. 1930-1978

The economic value of the reduction of mortality in this period was estimated to be worth $846 billion in 1975 dollars, while an independent estimate of biomedical research expenditures gave $86 billion. If we take only 20 percent, rather than 23 percent, as the contribution of biomedical research to mortality reduction, we obtain a net return to such research of $83 billion. This does not capture the improvement in the health status of the living. Clearly, the payoff to this type of research was as impressive as the payoff to technological progress in the Solow calculations.

Measurement Issues

The above mentioned studies are interesting not only because of the importance of the outcome, but also because they are so clearly of a multivariate character. Output growth and mortality decline are found to be only partially, though quantifiably, affected by innovative activities. Other influences, such as those of the stock of medical personnel or of capital, are included and evaluated by standard statistical The Denison-Solow results are based on regression estimates of production function parameters.methodology.

Politicians, lobbyists, and sometimes even scientists, when clamouring for increased support to research, only rarely pay attention to the existence of multiple causes. They prefer to dwell in a single bivariate world where single causes lead to unique outcomes.

These two studies also attempt to uncover whether innovative inputs have desirable impacts. Too often in public debate the attention is focused on only one or the other side of the innovative "function." Korea's economic growth is fast, therefore its R&D outlays are probably high; Phillips' research expenditures are large and increasing, therefore the electronics giants' profit growth must be strong. That the desirable economic outcome necessarily results from innovativeness, or that innovativeness necessarily leads to economic success has, however, no general proof. The connection, or more accurately, the partial connection between them must always be documented. This can be painful and difficult.

A hint of the inherent difficulties underlying the measurement issue is already implicit in the two studies mentioned. Why did Solow examine the growth of business output and not the growth of the total, that is business plus government output? We do not register productivity changes in the public sector because its output is measured by its inputs, with no changes in value added. Why did Vehorn et al. measure the decline in mortality when new drug therapies not only prolong life, but also make people enjoy better health while living, i.e., lower morbidity, the incidence of diseases? Simply because morbidity is a much more elusive concept than mortality.

Figure 1 offers some clues about what should be taken into account when trying to measure the degree of innovativeness, its determinants, and its consequences. These (and other) measurement aspects, which will be constantly alluded to throughout this volume, are not examined for their academic interest. They are sketched here in a first "fly-by" so that the problems facing adherents of policies to stimulate innovativeness can be glimpsed.

Click here to view Figure 1: A Representation of Determinants (a) and Consequences (b) of Innovativeness

The Input Side

Panel (a) of Figure 1 deals with what can be called the input side of innovativeness, as well as with its two constituent parts. Let us consider those two first. Since innovativeness was defined as consisting both of product/process innovation creation and of innovation adoption/diffusion, it is readily apparent that its degree or intensity cannot be measured directly. An index of innovativeness can only be an aggregate of apples and oranges: technological breakthroughs mixed with incremental process improvements and widely differing diffusion rates in a host of It will be seen later that similar difficulties beset an index of competitiveness. Nevertheless a start has been made on at least defining and counting innovations. The U.S. Small Business Administration has catalogued 8,074 innovations introduced in the United States in 1982. For an analysis of these data see Zoltan J. Acs and David B. Andretsch, "Innovation in Large and Small Firms," American Economic Review, September 1988, pp. 678-690.markets. The same reservations apply to potential indices of "just" innovation creation or diffusion.

And so it is not surprising that much of the discussion and investigation of innovativeness focuses on the imperfect indicator of research and development expenditures, or research intensity (R&D divided by sales revenue). At the same time the R&D indicator is also considered to be a prime determinant of innovation. It thus serves two purposes: as a proxy measure of innovativeness and as an input into the process of innovation. But as an input it is merely a necessary-certainly not a sufficient-contributor to industrial innovation.

Before we look at the other contributing elements let us pay closer attention to R&D. It is only over the last decade that we are obtaining ever stronger econometric evidence about the existence of so-called technological Jeffrey Bernstein, "The Structure of Canadian Interindustry R&D Spillovers and the Rates of Return to R&D," Journal of Industrial Economics (1989), v. 37, No. 3, pp. 315-328.spillovers. Technical knowledge gained through research spills over, without cost, to technologically similar industries. The recipient firms or industries need not be, and often are not, in the same country. As well, industrial R&D also relies on knowledge generated through basic research in universities and research institutes.

As has been repeatedly shown, successful industrial innovation is predominantly the result of skillful management. Management is the bridge between the laboratory and the market or the factory. Management evaluates a market or a process need, entrusts its researchers with meeting it, and guides the production specialists in delivering the new product or installing the new process at acceptable cost. The essence of innovation, states a recent (1990) research project outline of the Science Council of Canada, is the commercialization of technology rather than its mere creation.

Finally, a typical innovation requires a substantial investment. It must be financed not only during its gestation in the lab, but also in the market investigation phase, in the setting up of pilot runs, and production and distribution facilities, in advertising its launching to markets. Such an investment will only be incurred when business conditions-cyclical as well as political-are welcoming.

What, then, are the implications of panel (a) for the would-be dispensers of taxpayer funds to enhance innovativeness? The most obvious one is that the diffusion/adoption component must also be taken into account. The second implication is that since there is no satisfactory way to measure (no "index" for) either innovation creativity or diffusion, only some very indirect proxies can be pursued as objectives, such as R&D activity.

Yet the objective of increasing R&D effort suffers from two handicaps: not only is R&D but a proxy variable for innovation, it is also only one of several inputs into innovation creation.

We singled out business conditions as being crucial to the investment plans behind innovative products and processes. Such conditions are fundamentally influenced by the macroeconomic as well as "target-neutral" policies of the governments. The encouragement of a high level of savings and capital formation by sound fiscal and monetary policies, the provision of an adequate infrastructure of technical and managerial training, the absence of stultifying regulation are but some examples of what is meant by policies favourable to E.P. Neufeld, senior V.P. of the Royal Bank, before the Senate Subcommittee on Estimates, May 12, 1983.investment. Clearly such policies are much more exacting than just direct support to R&D.

The input factor "management" is particularly relevant at the level of the individual enterprise. Should only well-managed enterprises be the beneficiaries of taxpayer support to R&D? How do we predict the quality of management during the course of the supported innovation project? Can bureaucrats pick winners?

And if taxpayers are to subsidize one specific industry, they will inevitably contribute-because of spillovers-to the innovativeness of another, possibly foreign, sector. These are some of the issues that anybody considering how to increase innovativeness in a firm, industry, or country must grapple with.

It is not idle, even at this introductory stage, to insist that general framework policies which make for a sound economy, as well as managerial skills (especially in marketing), are as much responsible for industrial innovation as scientific research and development itself. Even NABST, the National Advisory Board on Science and Technology to the prime minister, seems to stress only one part of the picture, the technology part, in its 1991 normative scenario for a Canadian industrial policy, as evident in Figure 2.

Click here to view Figure 2: Routes to National Prosperity

The Output Side

Panel (b) of Figure 1 indicates the possible consequences of innovativeness and the output proxy measures of it. Again, no single satisfactory measure is available.

Patents, for instance, will not be present in an industrial sector (or in an economy) where innovative activities are mostly of the adoption kind, or where intellectual property cannot be adequately protected by them (eg. mining, petroleum refining). Apart from innovation, the balance of trade in a specific sector, i.e. the difference between exports and imports, will clearly be influenced by a whole host of factors, including, for instance, a sector's natural resource endowment. Total factor (capital plus labour) productivity will in part depend on the amount of capital available and on the training of labour.

Nowadays it is fashionable to subsume a sector's long-term health and prospects under the word "competitiveness." One could therefore also propose competitiveness as an aggregate index of the outcomes of innovativeness. The difficulties of constructing such an index, which typically consists of at least two components-trade performance and productivity growth-are obvious. But since at some very basic level the notion of competitiveness seems to be right and important, we shall pay further attention to it subsequently.

Nevertheless, since the c word is now so freely bandied about, not just in connection with a particular sector of industry but also in relation to a whole economy, it is advisable to quote here and now a well known observer of the economic scene, Paul Krugman of MIT:

Indeed, trade between countries is so much unlike competition between businesses that many economists regard the word "competitiveness," when applied to countries, as so misleading as to be essentially Paul A. Krugman, "Myths and Realities of U.S. Competitiveness," Science, November 8, 1991, p. 811.meaningless.

Is Canada Innovative and Competitive?

Having sketched some aspects necessary to the preliminary understanding of the policy issues regarding innovativeness, we now present some statistics which preoccupy so called decision makers in this area. A greater in-depth exploration is reserved for subsequent chapters.

The most infamous and notorious indicator of the alleged lack of innovativeness in Canada is the GERD/GDP ratio, or the gross expenditures on research and development divided by the gross domestic product. As Figure 3 shows, this presumed indicator of an economy's innovativeness-as signalled exclusively by research and development-shows Canada to be in the minor leagues, with a ratio of 1.33 percent in Later Canadian figures are virtually unchanged; international figures come with a delay of 3-4 years.1989.

Click here to view Figure 3: GERD/GDP, Approximate Percentages, 1989

This is the statistic most frequently used in debate by opposition parties (Progressive Conservatives before 1984, Liberals after, Socialists throughout), by the media, and by many science and technology interest groups to show that Canada lags behind in innovativeness, and so ultimately, in competitiveness. While we shall return at greater length to this very indicator, it is of interest at this point to present-in Figure 4-an intriguing relationship between the size of an economy and its outlay on R&D.

Click here to view Figure 4: Log of GERD as a Function of Log of GDP 1987, 19 countries

A regression line of "best fit" between the natural logarithms of GERD and GDP appears to have a slope statistically significantly greater than 1 (actually 1.16): the larger an economy, as measured by its domestic product, the higher a proportion of it is spent on research. If this exponential relationship could be explained on economic grounds, policy-makers' efforts to increase the GERD/GDP ratio would seem to lose even more of their The figure is taken from J.A.D. Holbrook, "The Influence of Scale Effects on International Comparisons of R&D Expenditures," Science and Public Policy, August 1991, pp. 259-262. The countries are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Iceland, Ireland, Italy, Japan, Netherlands, Norway, Spain, Sweden, Turkey, U.K., U.S.A. The economic explanation could be based on the assumption that it is easier to appropriate returns to innovation in a larger jurisdiction. See Bohumir Pazderka, "Determinants of the International Distribution of R&D Expenditures," Canadian Economics Association Meetings, June 1992.justification.

The next graph (Figure 5) purports to show a deterioration in Canada's competitiveness over the years 1980 to 1990. Competitiveness, just as the GERD/GDP ratio, is conceived of as an economy-wide measure. In this representation, offered by the Canadian Manufacturer's Association in 1990, an index composed equally of unit labour costs, wholesale prices, and international trade is pitted against a similar composite index for the other countries comprising the G-7 (US, UK, Japan, France, Germany, Italy). The aggregative steps taken in compiling the index are breathtaking and are compounded by the element of international comparison. If innovativeness is an important element contributing to competitiveness, then-on this reckoning-our presumably poor showing on one index implies a poor performance on another.

Click here to view Figure 5: The Canadian Competitiveness Index

The third measure of alleged shortfall in innovativeness is actually a small component of the previous one, but more sharply focused. It is the share of trade in high-technology products. Again, we shall return to this issue in greater detail later asking, for instance, why the definition of hi-tech does not include the nuclear reactor sector, or why market share is a decisive factor when in fact the whole market grows furiously. As Figure 6 indicates, Canada's share of OECD's exports of the so-called hi-tech products has declined slightly over the 28 years between 1963-1987. (At the same time, medium-research intensive products are doing very well over this period). But this argument must not pass unchallenged, even in this brief exposition. When Northern Telecom buys a component from its plant in Texas, this is registered as a Canadian import. If trade balances were calculated in terms of ownership, it would not. Is a geographical definition superior to one in ownership terms? It is well known in this context that U.S. corporate sourcing practices, that is, U.S. corporations buying from their own plants abroad, account for a substantial part of U.S. trade imbalances with Taiwan, Singapore, and South Korea.

Click here to view Figure 6: Canada's Share of OECD's Exports by Technology Type, 1963 to 1987

Figures 3 (GERD/GDP), 5 (Competitiveness), and 6 (Trade in Hi-Tech) thus express in a condensed way the malaise about Canadian innovativeness that has been part of this country's conventional wisdom for at least two decades.

Direct Government Support for Innovativeness

During the same time, a blanket accusation against successive federal governments states that not enough public funding is being committed to innovative activity in general, and to research and development in particular. The federal taxpayer supports research through various outlets: federal laboratories (such as in agriculture and environment), grants (to universities and industries) and tax abatements to private sector research performers. In a later chapter this topic will be explored at greater lengths. Here we show, in Table 2, perhaps the most important facet of taxpayer's generosity toward research. It is an international comparison of Canada's tax treatment of Jacek Warda, International Competitiveness of Canadian R&D Incentives: An Update, Report 55-90, Ottawa: The Conference Board of Canada, June 1990.R&D.

Click here to view Table 2: Comparison of B-Indexes in 10 Countries, 1989 and 1981

The table shows a so-called B-index rating of tax incentives available to research-performing corporations. It is defined as after-tax cost of $1 of R&D expenditure, divided by one minus the tax rate, or ATC/(1 - tax rate). Both in 1981 and in 1989 Canada ranked first in generosity, if the corporation was subject to the Quebec taxation regime. (If the B.C. corporate tax system were to apply, the least stimulative in Canada, then Canada's rank would decline to third place, if Ontario's, then to the second). This investigation was carried out by the Conference Board of Canada and financed by the federal department of Industry, Science and Technology. Its results indicate strongly that at least as far as the tax system (federal and provincial) is concerned, Canada is more than enough competitive.

Summing Up

In this chapter we have described a number of building blocks that should go into the construction of a book dealing with Canada's innovativeness and with policies aimed at improving it.

We stated that technological innovativeness consists of both the creation and the diffusion of new products and processes. We showed two examples-one of economic growth, the other of mortality decline-due to innovativeness. This to illustrate the general beneficial effects of it. Yet we cautioned that it is quite difficult, in most instances, to be precise about the role that innovativeness plays in such a complex phenomenon as economic prosperity. Definitional and measurement problems account for this: R&D is but a proxy measure of the larger concept, for instance, and it is only one factor in the multivariate causation of economic prosperity.

Figure 1 attempts to illustrate the argument and represents, in a simplified manner, the underpinning for this whole monograph. Innovativeness can only be represented indirectly by input (R&D) or output (total factor productivity, etc.) proxies. There are several other important inputs into innovativeness (management, etc.), and other outputs by which it can be assessed in part (patents, etc.). Sometimes several of these proxy outputs are brought together into an index of competitiveness.

Figure 1 led us to preliminary reflections on the usefulness of pinpoint industrial policies toward R&D, on the subsidization of a domestic sector which may contribute to the innovativeness of a rival foreign sector, and on other issues.

Finally, this chapter listed graphs and statistics supporting the primary concerns of Canadian media, politicians, and technology lobbies. These are the GERD/GDP ratio, the competitiveness index, and the hi-tech products trade balance. We noted that these concerns occur despite the most generous R&D taxation climate anywhere.

Looking Ahead

Future chapters will present a coherent view of the state of Canada's innovativeness and of the policies-past, present and proposed-designed to improve it.

The next chapter begins by defining industrial policy and by examining traditional (market failure) and its recent (strategic trade) conceptual foundations. Since much of the debate about industrial strategy centers around technological innovation, it will not be necessary to make a special effort to narrow it down to our own topic.

Having established the theoretical reasons for a public policy debate about innovativeness, we must next deepen our understanding of it. This will be undertaken in chapters 3 and 4. Chapter 3 will look at innovativeness from an economic perspective, while Chapter 4 will add a managerial perspective, especially as regards determinants of innovation other than R&D.

In chapter 5 we shall examine in some detail indicators of Canada's innovative performance and competitiveness. Chapter 6 will discuss industrial policies or governmental support for innovativeness in Canada, mainly under the rubrics of taxation, grants, patent policies, and government-performed research. In chapter 7 some foreign innovation policies and their outcomes will be described, with a view to comparing them with our own. The last chapter will summarize the descriptions and arguments presented in the book and will give some conclusions.


Chapter 2 Industrial Policy

Propositions about where and how market forces work poorly, however, cannot alone carry the policy discussion very far.

R.R. Richard R. Nelson, High-Technology Policies: A Five-Nation Comparison, Washington, D.C.: American Enterprise Institute, 1984, p. 5.Nelson

Definitions

WHILE THE EXPRESSION "INDUSTRIAL POLICY" or, interchangeably, "industrial strategy" has been around for at least three decades, the idea and its workings go back at least three centuries. In response to a request by the Economic Council of Canada Albert Breton formulated a solid definition of it in 1974. Three centuries before that, the finance minister of Louis XIV, J.B. Colbert, practiced it by promoting and sheltering from competition a number of then hi-tech industries, such as armaments, shipbuilding, and tapestry "...he proceeded to reconstruct commerce and industry according to the economic principles known as mercantilism. Utilizing protective tariffs, government control of industry and trade, and navigation laws, Colbert organized trading and colonization companies, established model factories and succeeded in extending French industry and trade... Coupled with the extravagance of the king, Colbert's programs drained the French economy." Funk & Wagnalls New Encyclopedia, New York: 1973, under Colbert, Jean-Baptiste.manufacture.

The Breton definition of industrial strategy starts with a notion that all goods and services, both intermediary and final, can be divided into several broad conceptual classes such that the total output of an economy, Q, in a given period can be represented by the sum of individual sectoral sub-outputs

Q = A + R + S + M + ...

where A stands for agriculture, R for resources, S for services and M for manufacturing or "industrial" and the three dots represent other possible output Albert Breton, A Conceptual Basis for an Industrial Strategy, Ottawa: Economic Council of Canada, 1974.classes.

For certain problems it could be appropriate to make a breakdown to several types of M-output such that, for instance,

M = M1 + M2 + M3 + ...

where M1 is high-technology, M2 medium- and M3 low-technology industrial sectors.

Suppose now that some socially optimal level of Q or M and of its components can be defined and is designated by asterisks:

Q* = A* + R* + S* + M* + ...

or M* = M*1 + M*2 + M*3 + ...

(To the economist a socially optimal level of output means that perfect competition prevails in all of the economy's markets and all externalities have been internalized-all sectors are at the peak of efficiency and no economic agent can be made better off by changes in the output's composition. To the politician a socially optimal output may not mean anything remotely resembling this abstruse concept. To her it may signify an optimal chance at re-election). Given these concepts the definition proposed by Breton is as follows:

An industrial strategy is an attempt to reduce the gap assumed to exist between the actual outputs of M-goods and the socially optimal level, between M and M*.

An important aspect of this view of industrial policy derives from its general equilibrium formulation: when the goals of the strategy are defined in terms of the size of a sectoral output, they are also defined in terms of output composition or of the relative size of sectors. An increase of M towards M* implies a change in one or more of the other sectoral outputs, such "The chronic excess demand for loans created by the administered below-market rate allowed the Ministry of Finance to engage in effective credit rationing to the point of guiding the largest banks to make loans to a specific industry or even to specific firms engaged in the investment race." Kozo Yamamura, "Caveat Emptor: The Industrial Policy of Japan" in Paul R. Krugman (ed.), Strategic Trade Policy and the New International Economics, Cambridge, MA: 1986, p. 172.as A or R.

While Breton's may be the most rigorous definition on record, we often encounter proposals that merely state certain grand "Therefore the Council recommends that: The government, in consultation with industry, set realistic goals for each industrial sector, work with industry to develop action programs, and publicize the goals and achievements." Report of the Advisory Council on Adjustment, Adjusting to Win, Canadian Manufacturers' Association, Toronto: March 1989, p. 80.objectives, frequently inherently contradictory and usually banal, intended somehow to guide government policies and corporate Gordon Ritchie, "Government Aid to Industry: A Public Sector Perspective," Canadian Public Administration, 26 (Spring 1983), pp. 37-46.decision-making.

Before we turn to some illustrations of industrial policy we must add to this section a description of what could be called the international trade version of industrial strategy, namely, strategic trade policy. We must do so because in the last half a dozen years this version, or offspring of industrial policy, has become more intensely discussed than its parent.

Perhaps the best way to indicate the intertwining of industrial and trade strategies is to cite from a recent US congressional study:

Japan and other Asian countries have combined numerous policy tools besides long-term government support for technology R&D to promote selected industries: preferential loans from government banks or banks that follow the government's lead; guaranteed purchases by governmental bodies for home-grown products (e.g., semiconductors for Nippon Telephone & Telegraph, supercomputers for government agencies); government-subsidized leasing companies making guaranteed purchases of advanced equipment and leasing them at preferential rates (e.g. robots, CNC machine tools); formal or informal barriers against imports, removed (or partly removed) only after the domestic industry has become a world-class competitor; strict limits on foreign investment in manufacturing; government negotiations for technology licenses on behalf of industry; government guidance (not always followed) to rationalize industries, scrap overcapacity, and encourage companies to get economies of scale by specializing in certain parts of an industry (e.g., machine tools).

This is industry cum trade policy on a comprehensive U.S. Congress, Office of Technology Assessment, Making Things Better: Competing in Manufacturing, OTA-ITE-433, Washington, D.C.: U.S. Government Printing Office, February 1990, p. 79.scale.

An approximate definition of strategic trade policy can be gleaned from a description of the three constituent words. Trade implies that the subject of the policy is movement of goods and services across borders and, increasingly, on a global scale. Strategic relates to the international interdependence of policy actions in an oligopolistic environment, that is, in markets dominated by only a few firms. Policy means here that governments are the main actors in this game. Each government takes into account some response by foreign firms or governments in calculating its own best course of Klaus Stegemann, "Policy Rivalry Among Industrial States: What Can We Learn from Models of Strategic Trade Policy," International Organization, 43 (Winter 1989), pp. 73-100.action.

Finally, we should make a distinction between the industrial and trade policies defined above and "general framework" or "target-neutral" policies. These latter policies do not "target" specific industries (such as hi-tech or bankrupt sectors) or specific activities (such as technology); they aim to provide the economy with general conditions favourable to growth. Such policies may imply, if they are to stimulate innovation-based growth, fiscal restraint, conservative taxation and limited intervention.

This is how a recent report of the European Industrial Research Association describes Switzerland's posture toward innovation:

The main task of the government is to support innovation through the creation, maintenance and support of the infrastructure for education and research as well as to generate a climate favourable to innovation. The latter implies stable and reliable economic conditions. Another important duty of governments is to generate the necessary social acceptance for the role of Switzerland as an industrial European Industrial Research Association, Impact of Government Policies on Industrial R&D, Working Group Report No. 41, Paris: 1990.nation.

While the rest of this chapter focuses on industrial and trade policies that support innovativeness, we would be ill advised to lose sight of the overriding importance of framework policy. Examining high-technology oriented policies in the five leading industrial countries, Richard Nelson asked the crucial question in this context: to what extent does strength in these (hi-tech) industries flow from general economic strength rather than the other way Richard R. Nelson, High-Technology Policies, op. cit., p. 5.around?

None of the above attempts at definition should blind us to the fact that the divisions between industrial, trade, or framework policies are merely approximate. In an address to the 1971 meetings of the British Association for the Advancement of Science, its president, Sir Alec Cairncross, opined that innovation is a form of investment and when the economic and social climate is uncongenial to investment, it naturally discourages innovation, Sir Alec Cairncross, "Government and Innovation" in G.D.N. Worswick (ed.), Uses of Economics, Oxford: Basil Blackwell, 1972, p. 19.too.

Rationales for Public Intervention: Industrial Policy

The interference of governments with market processes dates from time immemorial and is the rule rather than the exception whether in advanced economies, developing, or socialist ones. The justification for such intervention ranges from religious to social to economic concerns. Here we concentrate on the economic rationales underlying government activities that are aimed at influencing economic agents (firms and investors) to act in certain ways that would lead to greater innovativeness.

Presumably these agents need to be influenced, by stick and carrot, to do things which they would not, if left to respond to market pressures. 0ur specific case assumes that the market detracts firms from some optimal or merely desired level of innovative behaviour such that, for instance, the share of high-technology product output in total manufacturing output is less than it should be:

M*1 (hi-tech) > M1 (hi-tech).

In that sense markets are said to "fail": there is some "flaw" in the nature of privately organized economic activities.

But how do we know or how can we estimate that

M*1 ¹ M1?

To this there are several possible answers. One, for instance, looks to comparisons of industrial output categories with foreign countries. Another finds its inspiration in arguments presented to elected (and non-elected) officials by special-interest groups. The one answer that is generally accepted in economic literature, at least as a starting point, explains the likely reasons why unhampered market forces fall short of driving firms, and so industries, to optimal levels of output. The most frequently mentioned reasons, over the last quarter-century, are the following ones.

Market Failure Due to Externalities

Firms producing new technology may not be able to exclude others in society (firms or final consumers) from obtaining at lower than full cost the benefits of the new technology. When a new product is sold at a uniform price to all comers (the standard occurrence in our markets and under our legal system), some customers who would have been willing to pay a higher price for the product than the one charged reap what is called a consumer surplus. If the product is an input into further production processes (e.g., a computing device sold to banks) this consumer surplus may show up in a measurable way, as an increase in the productivity of the customer industry. Had the innovator been able, through perfect price discrimination-charging a price fitted to each customer's eagerness-to appropriate all of the consumer surplus he created, he would have reaped the "full" returns to his In a well-known realistic example Griliches shows how a patent-protected innovator can increase his profit from $25,000 to $500,000 if he is able to appropriate, by perfect discrimination, all of the consumer surplus. See Zvi Griliches, "Issues in Assessing the Contribution of Research and Development to Productivity Growth," Bell Journal of Economics, Spring 1979, p. 98.venture.

Imagine now an industry, such as the semiconductor sector, in which patents are difficult to enforce and technical secrets hard to keep since scientists and engineers are highly mobile between individual firms-some of which may be foreign-owned. The protection of the rights of ownership to the intangible knowledge assets is difficult under such circumstances. Because the transit of the technological information is not effected through a voluntary value-for-value exchange for a price, such as must be the case in a market transaction, we speak of market failure.

Here, as in the previous case, we are in the presence of an externality springing from imperfect or unenforceable property rights. In both instances a positive or beneficial externality is in existence: an unintended outcome of one firm's (innovative) activity redounds to the benefit of customers or competing Dennis Mueller, Public Choice II, Cambridge: Cambridge University Press, 1989, ch. 2.firms.

Such a beneficial externality conferred upon customers is designated in economic jargon as a productivity spillover. When it is received by competitors it is called a knowledge or information spillover. While increasing the welfare of the economy, such spillovers obviously detract from the income of the innovator who is unable to appropriate fully the economic rewards of his new product or process. Spillover-created inappropriability is thus believed to lead to a lower level of innovation activity than would be experienced in markets which do not "fail": entrepreneurs are reluctant to invest as fully in ventures in which their rights are not totally secure.

The logical argument that intellectual property rights are often unenforceable and that this may depress industrial research and associated innovation activities leads then to the recommendation that the gap between private innovation benefit and its private cost be closed by a subsidy. Provided, of course, that the subsidy does not exceed the external benefits created by the innovation in question. We shall return to this basic, all-important argument for research subsidization in a more technical manner in a subsequent chapter. Here we should pause and ask to what extent the general expectation of inappropriability and its consequences is justified.

First then, do we have empirical evidence of spillovers? Research in this area has been accelerating over the last decade and appears to show the existence of spillovers, both between firms within an industry, and between industries-the inter-industry spillovers occurring even on an international scale.

The first to look into productivity spillovers, or the effects of innovative efforts of supply industries upon downstream customers, was Terleckyj as long ago as Nestor E. Terleckyj, Sources of Productivity Advance, Unpublished Ph.D. dissertation, Columbia University, 1960.1960. He found that most of the productivity increases assignable to innovations could only be detected in customer industries. His findings were repeatedly confirmed both with Canada-wide and Quebec H.H. Postner and L. Wesa, Canadian Productivity Growth, Ottawa: Economic Council of Canada, 1983. Petr Hanel, J.F. Angers and M. Cloutier, L'effet des dépenses en R&D sur la croissance de la productivité, Québec: Ministère de la Science, 1986.data. Box 2 shows some of the empirical findings. This type of spillover, or external effect, can therefore be taken for granted. While it contributes the necessary underpinning for subsidization advocates, it is not by itself sufficient. For that, the subsidiser must be able to possess information not easily come by, and must not be influenced by personal considerations. We shall elaborate on this in a later chapter.

Click here to view Box 2: Productivity Spillovers in U.S. and Canadian Industries

Knowledge or information spillovers are the much "hotter" topic, attracting more recent attention. The subsidy justification is here based on the notion that the new technological knowledge generated by one firm seeps out to other firms in its industry (intra-industry spillovers) or in other industries (inter-industry spillovers). But there is no reason to expect such seepage to go one way only. Clearly, if there are externalities in this situation, they are reciprocal: each party both receives from and emits an externality to the other Robin W. Boadway and David E. Wildasin, Public Sector Economics, 2nd. Edition, Toronto: Little, Brown, 1984, p. 105.party. While the reception may go unacknowledged by the subsidy candidate, it cannot be neglected by the subsidiser who dispenses taxpayer funds.

Just as in the instance of productivity spillovers, the presence of information spillovers may be detected by productivity or cost Jeffrey I. Bernstein, "Costs of Production, Intra- and Interindustry Spillovers: Canadian Evidence," Canadian Journal of Economics, 21, May 1988, pp. 324-47.studies:

c = C(y,w,S)

where c is the firm's cost of production, y the vector of outputs, w the vector of factor prices and S a vector of spillover variables. (Among factors or inputs is also own R&D capital). The intra-industry spillovers can be defined as a function of the sum of the R&D capital stocks of all rival firms in the industry. Similarly, the inter-industry spillover-causing variable is defined as the sum of the R&D capital stocks for all other industries in the sample.

This type of a model was estimated by Bernstein with Canadian data from corporations in seven two-digit SIC industries from 1978 to 1981. Spillovers showed up as exerting a statistically significant downward pressure on the costs of the "receiving" corporations. Bernstein's investigations, while subject to econometric reservations, go much further than simply showing spillovers, intra- or interindustry. They also lead to estimates of the "wedge" between private and social returns to R&D, that is to the difference between what a typical firm in a given industry can expect to reap from its investment in research and what the returns on this investment are to other firms and A lengthier discussion of social rates of return is found in Chapter 4. industries. In another study Bernstein assigns in pinpoint detail the beneficial (i.e. cost reducing) spillovers received by an industry to the various "spilling" Bernstein, "The Structure of Canadian Inter-industry Spillovers, and the Rates of Return to R&D," Journal of Industrial Economics, 37, March 1989, pp. 315-28.industries. Finally, in certain industries spillovers act to depress (i.e., make less necessary) the receiving firms' investment in R&D, while in others they Irene Henriques, Four Essays on Research and Development and Spillovers, unpublished doctoral dissertation, Queen's University, Kingston, Ontario, September 1990.stimulate it.

Bernstein's and other investigators' work has several intriguing implications. One is that some industries are more senders than receivers of spillover externalities-and perhaps, therefore, ought to be singled out for support. The other is that, as we would expect from an already much older literature on oligopolistic rivalry, firms respond to their competitors' innovative activities by increasing their own.

In that sense, then, while spillovers may exist, they do not lead necessarily to underinvestment in research. Indeed, there is an ancient and perhaps ill-substantiated grudge against the pharmaceutical industry which maintains that there is too much "inventing around and so too much R&D W. Duncan Reekie, "Price and Quality Competition in the United States Drug Industry," Journal of Industrial Economics, 26, March 1978, pp. 223-37.investment."

Furthermore, we now have evidence of cross-border spillovers between industries. Mohnen finds that foreign (US, Japan, France, Germany, UK) R&D spillovers have played a major role both in the growth and in the slowdown of total factor productivity in Canadian manufacturing between 1969 and 1982. Their cost-reducing effect outweighs the own-R&D expenditure Pierre Mohnen, The Impact of Foreign R&D on Canadian Manufacturing Total Factor Productivity Growth, Centre de recherche sur les politiques économiques, Université de Québec à Montréal, July 1990.effect. They tend to stimulate-are complementary to-Canadian R&D outlays.

If there is a substantial spillover to Canada from abroad, it is logical to expect some technological spillovers from Canadian industries outward. The basic argument for subsidization of innovative activity is the inappropriability of full economic returns to it by the firm due to spillovers outward. But if much of the spilling consumer surplus flows to foreign customers, the claim on the purse of the Canadian taxpayer is "It appears likely that the new international economic environment has reduced the ability of any one national economy to appropriate the economic returns from basic research performed within its boundaries." David C. Mowery and Nathan Rosenberg, Technology and the Pursuit of Economic Growth, Cambridge: Cambridge University Press, 1989, p. 292.weakened.

Lastly, there is still some scepticism about the measurement of spillovers and about their intensity. One of the leading US econometricians, Zvi Griliches, opines Zvi Griliches, "Productivity Puzzles and R&D: Another Nonexplanation," Journal of Economic Perspectives, v. 2, Fall 1988, 18-19. See also David M. Levy and Nestor E. Terleckyj, "The Problem of Identifying Returns to R&D in an Industry," Managerial and Decision Economics, Special Issue, 1989, pp. 43-49.that:

Recently there have been several interesting attempts to capture the impact of such spillovers and to model their spread but they have not yielded a convincing estimate of their global impact. Our current understanding of this whole process is still seriously flawed. Moreover, without a major revision and extension of the national income accounts and the development of new data and methods for tracing the flow of ideas from one sector to another, researchers are unlikely to do much better in the near future.

The intensity or pervasiveness of spillovers of new technological knowledge is doubted by some prominent students of the subject. They point out that the transfer and utilization of technical information is a costly undertaking and one that requires a technological receptivity that is based on adequate research-based knowledge of the relevant subject area. Thus Cohen and Levinthal: "...our analysis suggests a lesson for technology policy...it implies that the negative incentive effects of spillovers and, thus, the benefits of policies designed to mitigate these effects, may not be as great as W.M. Cohen and D.A. Levinthal, "Innovation and Learning: The Two Faces of R&D," Economic Journal, 1989, p. 594.supposed."

We end up by acknowledging that the implications of information spillovers, should their widespread existence be confirmed, are far-reaching. Nonetheless, we consider them, as regards industrial policy intervention, quite ambiguous.

Market Failure Due to Other Factors

Higher differential risk or uncertainty in an industrial subsector may arise for several reasons, such as the absence of an insurance market or fewer opportunities for portfolio diversification. The alleged comparative (to other subsectors) affliction of the hi-tech manufacturing subsector could, in principle, be alleviated by public underwriting of risky projects or by direct investment in venture stocks.

The presence of competitive imperfections in the form of cartels, monopolies-whether due to increasing returns to scale or not-and oligopolies, or of similar phenomena on the buyers' side, leads to inefficient market outcomes. Where prices do not indicate relative scarcities, market forces will not allocate resources efficiently and output level will be below optimum. However, it is often maintained that the existence of monopoly (and what are patent rights if not a temporary monopoly) or at least its promise, favours innovative behaviour. In general, monopoly has an ancient claim on public intervention. Usually, however, this type of intervention goes under the name of competition, rather than industrial policy. It is, of course, not only possible but quite likely that the source of much monopolization is government policy past or present, whether it encourages a cartel's formation (e.g. in uranium) or protects its existence (e.g., marketing boards) or indeed launches a monopolistic enterprise itself (Canada Post Corporation).

Public policy itself, then, may be an "outside of the market" force which leads to "market failure." It need not, of course, result in monopoly, but rather more generally in the distortion of the sector's output level. A clear example is the 1980 National Energy Program (NEP) of the federal government which pursued the two undoubtedly inconsistent goals of Canadianization and self-sufficiency.

Finally, it is sometimes alleged that the minimum efficient scale (MES) of innovative operations, such as laboratories, is so large that only one or a few enterprises can afford to undertake them.

Noll provides a masterful though somewhat aging exposé of the theoretical issues and of the empirical importance of these types of market Roger Noll, "Government Policy and Technological Innovation," in K.A. Stroetmann (ed.), Innovation, Economic Change and Technology Policies, Basel: Birkhauser, 1977.failure. Bletschacher and Klodt, meanwhile, offer a similar, though much broader up-to-date Georg Bletschacher and Henning Klodt, "Braucht Europa eine neue Industriepolitik," Kieler Discussion Paper 177, Institut fur Welwirtschaft, December 1991.survey.

Rationales for Public Intervention: Strategic Trade Policy

We have noted that an active, innovative behaviour by one firm or industry may stimulate, unwittingly, a similar increase in such behaviour in competing firms or industries, at home or abroad. But to talk of spillovers in such a case is somewhat awkward. Better perhaps to refer to rivalry, or strategic response.

Strategic trade policy justifications start from the observation, now widely accepted, that a country's comparative advantage in trade can change as a consequence of private economic activity or even of government These paragraphs rely heavily on Stegemann, op. cit. footnote 8.policy. Not just natural ability or factor proportions, but entrepreneurial initiative using innovative products or processes to be first to market and to exploit learning curve economies appears to confer often lasting advantage as well. Similarly, governmental support to technical education can enhance the natural ability of a country's inhabitants.

Two models of strategic trade policy demonstrate the possibility that a government can improve national welfare by "shifting profits" from foreign to domestic firms. The first, originally proposed by Brander and Spencer in 1981, is nothing else but a game played in "third" countries by duopolists, each originating in a different A more closely fitting reference here dates from 1985: James A. Brander and Barbara J. Spencer, "Export Subsidies and International Market Share Rivalry," Journal of International Economics, 18, February 1985, pp. 83-100.country. A particular country's government assures its exporter, whom it wishes to set up as a "Stackelberg leader," of subsidies supporting its sales in third countries, where it is in competition with the other world-industry The most frequent example offered is that of Boeing versus Airbus.duopolist. "Stackelberg leadership" shifts profits from the "follower" to the leader and so to the leader's A pithy description of game theories and Stackelberg outcomes is given in Jack Hirschleifer, Price Theory and Applications, 4th ed., Englewood Cliffs, N.J.: Prentice Hall, 1988, ch. 10.country. Why would, however, the Stackelberg duopolist not be willing to assume the burden of temporarily lower prices in his export markets given the vast profits reckoning from his ultimate leadership?

The export subsidy granted by the government presumably establishes a credibility that even if duopolist A were to be met by B's price reductions he would persist and would not retreat from his attempt at leadership. Thus credibility emerges as the only reason for governmental strategic trade policy; when two opposing governments get into action, losses are likely to persist on both Despite booming sales of the Airbus, the loans subsidizing it appear nowhere near their payback, according to Bletschacher and Klodt, op. cit.sides.

The second model upon which a rationale for trade policy can rest in principle was proposed by Paul R. Krugman, "New Theories of Trade Among Industrial Countries," American Economic Review, v. 73, May 1983, pp. 343-7.Krugman. It is the familiar infant-industry argument for initial government protection, enriched by a strategic dimension. If a duopolist is given protection in his domestic market, he receives a scale of production advantage-which may be of the learning curve type-over a foreign rival. This scale advantage then leads to lower costs and so to higher market shares even in third, unprotected markets. By the same token the rival duopolist's costs increase due to lower sales and production. The strategic reason for governmental intervention is that of credibility, as in the preceding model: it would not be credible for a duopolist to attempt profitable expansion of export sales on his own.

In a review of these newly-minted theories of international trade Stegemann puts the important question: do models of strategic trade policy provide guidance to government action?

In Stegemann's opinion the revolution attempting to change the traditional free-trade non-interventionist thinking of the economics profession is ebbing rapidly. Also crumbling with it are the theoretical and empirical supports for strategic trade policy:

A combination of reasons seems to have been responsible: the realization that the apparent policy implications of models of strategic trade policy are highly sensitive to changes in the special assumptions of these models; the difficulty of identifying real-world situations in which the special assumptions apply; the recognition that the costs of implementing strategic trade policies might easily exceed the benefits, even if appropriate "target" industries could be identified; and the apprehension that economic theory was becoming a supplier of intellectual ammunition for powerful forces that favour protection of particular sectors for the "wrong" Stegemann, op. cit., p. 90.reasons.

A less pronounced, but still sceptical view of the efficacy of strategic trade policies, is offered in a later paper by Paul A. Krugman, "Myths and Realities of U.S. Competitiveness," Science, November 8, 1991, pp. 811-815.Krugman.

It will suffice to single out two instances to illustrate this list of caveats. One of the assumptions underlying profit shifting is that the domestic firm selected for government support is domestically owned. As the intertwining of ownership across borders increases and joint ventures proliferate, particularly in risky hi-tech sectors which are the usual target for intervention, this assumption becomes increasingly tenuous.

The other shortcoming which merits mention is the partial equilibrium perspective inherent in governmental support to a single industry. As was stressed in the discussion of the Breton definition of industrial policy, stimulating the output of one sector by public intervention will generally have a negative effect on the output of other sectors. The implication of the one-industry support was tested by Dixit and Grossman who modelled several oligopolistic industries dependent on a common resource in fixed supply, such as Avinash K. Dixit and Gene M. Grossman, "Targeted Export Promotion with Several Oligopolistic Industries," Journal of International Economics, 21, November 1986, pp. 233-49.scientists. When trade intervention attempts to shift profits by stimulating a domestic firm to increase output, it reduces the profitability of the other industries, leading them to contract.

Stegemann concludes his survey of the strategic trade policy arena by stating that economists will continue to resist the popular presumption that a policy enhances national welfare if it raises the market share of domestic producers at the expense of foreign Stegemann, op. cit., p. 99.firms. They will keep throwing cold water on mercantilist ideas and will lay bare the group interests served by a particular policy at the expense of other groups in society.

As Grossman put it recently in an important paper, ". . . strategic interventions seek gain at the expense of trade partners, and so invite retaliation . . . but the ultimate goal ought to be a co-operative outcome in which all parties desist from pursuit of strategic Gene M. Grossman, "Promoting New Industrial Activities: A Survey of Recent Arguments and Evidence," Paris: OECD Economic Studies, No. 14, Spring 1990, p. 19.gains." Yet, as Stegemann states, the notion of the state having the responsibility to "shape" the comparative advantages of its industry in competition with other states is politically powerful. An indication of how powerful it is will be furnished in the next section.

The Debate About and the State of Industrial Policy in Canada

The everlasting debate about this topic in media and in political forums testifies to the strength of the idea that the state should intervene in a pinpoint manner in market processes in order to enhance this or that firm's or industry's It is not easy and probably not fruitful to go on maintaining a distinction between industrial and strategic trade policies. The distinction served its purpose in uncovering their intellectual roots.competitiveness. And so does the imposition of actual industrial policies upon large sectors of the economy despite their frequent and manifest failure.

One of the more animated recent debates about industrial policy took place in Ontario. The Liberal provincial government of David Peterson, now consigned to the dustbin of history, was very much attuned to the industrial strategy propositions of Harvard's Robert Reich, the liberal economic strategist behind the unsuccessful presidential candidate, Michael Dukakis, and the more successful Bill Clinton.

In November 1987 during the annual first ministers' conference, Premier Peterson proposed "that in order to increase productivity and international competitiveness Canadian governments should provide leadership in committing (sic) to R&D as a national Government of Ontario, A Commitment to Research and Development: An Action Plan, 2nd ed., Toronto, January 27, 1988, executive summary.priority. This commitment can best be expressed in the form of a national R&D target . . . . This action plan provides a policy planning framework to achieve a 2.5 percent R&D target (of gross domestic product) within ten years." The two principal reasons given for increased government support of private sector research are spillovers of the information kind and exceptional Op. cit., p. 15.riskiness.

This proposal and a budgetary commitment of $1 As is usual in such commitments, the funds were to be doled out over a period of many years. On the assumption that $200 million was to be paid out each year and discounting at a modest 12 percent, the present value of the billion shrinks to about $750 million. Furthermore, the Technology Fund, announced when the Premier's Council was created, took half of its budget from existing funds. Also, more than a third of the money is not hard cash but tax credits for companies doing research and development (Globe and Mail, January 14, 1991, p. A6). The (mid-?) 1990 annual report of the Ontario Technology Fund lists as actual cumulative expenditures a total of $155 million, covering the fiscal years 1986/87 to and including 1989/90.billion to a so-called Technology Fund was in part the result of deliberations of the so-called Premier's Council, a "multipartite" advisory body chaired by the premier and composed of certain cabinet ministers and "leaders" of business, labour and academic communities. The Council, established in April 1986, was given the mandate to "steer Ontario into the forefront of economic leadership and technological innovation." Sometime in 1989 (the three volumes are not dated) the Premier's Council issued a massive report calling for large-scale government intervention within the scope of Ontario's provincial Competing in the New Global Economy, Toronto, 3 volumes.powers. This report represents one of the latest and probably the most detailed advocacy of industrial policies in Canada ever. It is also a compendium of not necessarily unbiased bits of information about industrial policies and assistance abroad and in One year later, in January 1990, the report was followed by a much-trumpeted conference in Toronto sponsored by the Council, on the theme of global competitiveness. Or rather, more precisely, on how to devise a strategy for Canada's and Ontario's economic future in a globally competitive world. (Globe and Mail, January 16 and 17, 1990, 32 both).Canada.

The industrial policy and trade strategy recommended by the Council in its Competing in the New Global Economy follows what are by now standard lines in this "business," though perhaps with somewhat better insight:

1.Assisting restructuring in the industries by

a)offering potential investors in mid-size, Ontario-based, export -oriented firms tax initiatives in new share issues; the firms to be on a government-approved list

b)helping in worker adjustment

2.Investing in high growth and emerging industries

a)yet more R&D tax incentives

b)a strategic procurement plan for Ontario government and Ontario Hydro

c)a risk-sharing fund for export-oriented firms

d)refocusing the (public-fund-dispensing) Ontario Development Corporation to provide assistance only to businesses in manufacturing and exporting sectors

3.Improving the entrepreneurial climate for start-up companies in the same sectors

4.Meeting the science and technology imperative by

a)redirecting government research to industry

b)the establishment of seven Centres of Excellence, comprised of university and industry participants, to encourage transfer and diffusion of technology into industry.

And so on ...

These proposed policies are listed here merely as an illustration of typical proposed industrial strategy Note the persistent insistence on export orientation as a requirement of eligibility. This is probably the only avenue toward trade policy that a constitutionally limited provincial government can open.initiatives. Only 4b, the establishment of university-industry Centres of Excellence (materials research, information technology, groundwater research etc.) has been achieved, at the cost of $200 million over five years. The general evanescence of industrial strategy proposals and the impermanence of actual policy implementations in Canada-and elsewhere-is due to the similarly impermanent nature of democratically elected As an example, the Ontario government introduced a bill in 1988 for a so-called R&D tax super-allowance. The bill died with the demise, in the summer of 1990, of the Peterson administration. It was reintroduced in the legislature in December 1990. The Council on Technology is being recreated and its mandate broadened under the name of the Council on the Economy and Quality of Life, according to the Globe and Mail of January 24, 1991, p. A6.governments. There is generally less than meets the eye in industrial policies, proposed and actual. It is thus more fun and probably more to the point to argue about them than to describe them. It is therefore our forecast that the Ontario NDP government's industrial policy, unveiled in July 1992, will wither on the vine as Globe and Mail, July 29, 1992, p. 2.well.

For a fairly thorough description of some industrial policies in the USA, Germany, Sweden, France and Japan, and a good listing of Canada's and Quebec's innovation-oriented initiatives, Competing is a briefly useful compendium. A more global perspective is offered in the OECD's yearly Industrial Policy in OECD countries. For a truly analytical and objective examination of such policies abroad one needs to go to academic publications. This will be done in a later chapter.

But before we close this section we do offer a summary description of the most directly relevant federal subsidy or grant policies vis-à-vis innovation, circa 1990. The reader should remember, however, that much of this description will be, or is by now, out of date. To describe, analyze, or criticize innovation policies in Canada is to shoot at a moving target or to play with a kaleidoscope.

The n-th time freshly reorganized and renamed Ministry of Industry, Science and Technology Canada is "charged with ensuring Canada's international competitiveness through a strong, continuing integration of scientific, technological and industrial strategies and This and subsequent information is garnered from Department of Finance, Canada Budget Estimates, Part 3, 1990-91, Ottawa, February (?), 1990.activities." Since about 1987 most of the emphasis appears to be given to S(cience) and T(echnology). The ministry coordinates all of the government's S&T activities, it has an assistant deputy minister for "science advocacy" (e.g. National S&T weeks) and, of course, it hands out subsidies to an alphabet soup of S&T-related undertakings.

Table 3 gives some figures about such grants. The second largest subsidy, of about $45 million in fiscal year 1990/91, is to the IRDP (Industrial and Regional Development Program) action which is now wound down. While this program had a substantial component of innovation support, it suffered, at least initially, from some confusion. This is illustrated in Box 3.

Click here to view Table 3: Selected Grants by Industry, Science, Technology Canada

Click here to view Box 3: IRDP, or How a Government Policy is Implemented

By far the largest recipient of R&D (and other-type) grants is DIPP, the Defence Industry Productivity Program, with over $235 million in fiscal year 1990/91. It funds, among other things, research on aircraft engines at Pratt & Whitney in Quebec. Since it is primarily oriented toward exports, it lacks the "consumer surplus" justification: the spillover benefits to customers go abroad.

If there is any major tendency to be discerned in the helter-skelter collection of industrial mission strategies carried out by ISTC, it is away from pinpoint help to individual companies' R&D. It is oriented more toward diffusion and adoption (TOP-technology outreach programs, InnovAction-Centres of Excellence). It also finances, along vaguely Japanese lines, "precompetitive" research carried out by industry research consortia (strategic technologies programming, STP).

On the whole, however, basically, fundamentally-no matter how we put it-the overall target of Canadian and provincial innovation policies is the support of R&D. This is the tenor of most political pronouncements and the effect of most "strategies." In this context it is important to note that there is a major shift, on the federal scene, away from direct subsidy to business (most of which is in the control of ISTC) to tax credits to stimulate R&D outlays. This highly relevant observation will be documented and discussed at greater length in a subsequent chapter.

This concentration of attention on R&D is almost certainly a wrong posture on the part of the elected representatives. Let us quote from a recent, heavily data-based article in the authoritative Research Policy:

The above characteristics of R&D businesses can lead to policy implications. The most important implication for policy decisions is the demonstration that R&D is not an isolated decision. Heavy R&D spending is usually associated with a suitable competitive position, with related investments in marketing and production, and with policies affecting the product line, the human resource function and export J. Zif, D. McCarthy, A. Israeli, "Characteristics of Businesses with High R&D Investment," Research Policy, 19, 1990, pp. 435-45.activity.

Recall here Figure 1 which stressed the complementary nature of "management" and research, business conditions and research, and so on. Such complementary behaviour is now widely recognized in professional, bureaucratic, consulting and academic circles. Advocates of industrial innovation policies, such as the Ontario Premier's Council, propose therefore a much wider net of support to prospective "winners," stretching from credit facilities, to specialized labour training, to subsidies and tax breaks. But with such ambitious and wide-ranging interventions proposed, the implementation of prescribed industrial policies comes to the fore.

The Difficulty of Implementation

The fundamental reason for directing taxpayer generosity toward innovative activities in the private sector-for undertaking an industrial innovation policy-is the presumed existence of externalities. As was explained, the assumption is that the social product of innovation is greater than the private one. The difference between the two, it is thought, accounts for the less than socially desirable level of private investment in innovation (read: R&D). The gap should be bridged by some form of subsidy and so overcome the "inappropriability" deterrent.

The gap must be measured, however, or at least an effort made to estimate it. The gap is largest when it comes to basic research, smallest in relation to development (the D in R&D). In Canada, it will be shown later, federal support to industry does not show a systematic preference for this ordering.

While it may be the case that it takes more than forty years for the simple economic idea regarding appropriability to make its way into the decision bunkers of hi-tech subsidizers, we should look at the even simpler notion of "picking winners." This seems to be a substantial part of any industrial policy, and so of the innovation kind as well.

The idea is to put "your money," that is to say funds extracted from the unfortunate taxpayers, on firms or industrial sectors that promise to contribute to the growth of the economy and employment, but cannot do so without assistance. These may be "threshold firms," hi-tech firms, export-led sectors, whatever. The essential issue is to forecast which ones will be winners.

For this forecast government agencies will need to possess quite detailed information about how the economy works, what foreign competition will be up to, and how the targeted firm performs. Criticizing a 1982 Science Council of Canada proposed initiative to subsidize "threshold firms," that is firms about to take off successfully, Watson pointed out that if the Council has heard about such firms, so has the stock William G. Watson, "It's Still Not Time for an Industrial Strategy," Canadian Public Policy, 10, No. 2, p. 207; Guy Steed, Threshold Firms: Backing Canada's Winners, Ottawa: Science Council of Canada, 1983.market. And if the stock market, with its strong profit motive, has not received such information, why would we expect government bureaucracy to have unearthed it? As Watson further pointed out, the Science Council's study also revealed that several of the key threshold firms were on-or through-the threshold of receivership (in 1982).

The difficulty of choosing winners is forcefully illustrated in the Ontario Premier's Council proposal to back threshold Ontario Premier's Council, op. cit., volume 1, ch. 6.companies. The Council reveals a list of such companies in a table reproduced here as Table 4. The table lists Ontario Threshold Companies in 1987. On our count, by the end of 1990 at least 9 of the 25 companies had either gone into receivership, or had experienced grave difficulties, or had been sold to foreign interests.

Click here to view Table 4: Examples of Ontario Threshold Companies 1987

The failure to pick winners, to establish successful national champions or to see technology-based megaprojects to completion is also well-documented abroad and will be discussed Dirk de Vos, Governments and Microelectronics: The European Experience, Ottawa: Science Council of Canada Background Study No. 49, March 1983 and Henning Klodt, Wettlauf um die Zukunft, Tuebingen: J.C.B. Mohr, 1987.later. As Richard Nelson, one of the best known American economists in this field puts it:

A policy whereby government officials themselves try to identify projects that will be winners in a commercial market competition is always seductive, but the evidence, from our studies and others, suggests that such strategy is to be Richard R. Nelson and Richard N. Langlois, "Industrial Innovation Policy: Lessons from American History," Science, February 1983, p. 18.avoided.

And Davis et. al., in a closely reasoned analysis of U.K. state aid to industry, provide statistical documentation of the divergence between official industrial policy objectives and those actually Howard Davis, Dorward, Drive and Topple, "State Aid and Industrial Characteristics," Applied Economics, December 1980, pp. 413-28.pursued. In regard to R&D industrial subsidies it would appear that the principal policy goal was to "pick winners," that is, to give aid to those industries already favoured by the market as indicated by profitability and employment. The implicit objective revealed by the correlational analysis of disbursements and industry characteristics in 33 sectors receiving research aid in 1975 was the support of relatively unprofitable and high unemployment industries-an objective appearing to fly in the face of judgement by the market mechanism.

The Difficulty with Motivation

We have already intimated, with the help of Box 2, that the objectives of implementers of industrial policy may not match the objectives of the politicians. Clearly the objectives of politicians may not, in turn, match the interests of the economy as a whole.And so the remedies suggested for market failure, and their implementation, may lead to nonmarket failure. That type of malfunctioning of the government redistributive mechanism is at heart due to differing motivations: the interests of the policy makers and their civil servants do not coincide with those of the public. "Policy formulation properly requires that the realized inadequacies of market outcomes be compared with the potential inadequacies of nonmarket efforts to ameliorate Charles Wolf, Jr., "A Theory of Non-market Failure," Journal of Law and Economics, April 1979, pp. 107-139.them." This statement by Wolf reflects the accumulating experience with ill-advised or ill-executed governmental attempts to improve upon the allegedly inadequate functioning of sectoral markets; it also echoes a substantial amount of theoretical and empirical work documenting the logic of nonmarket failure.

The theory of nonmarket failure observes that it is frequently impossible to organize nonmarket (e.g., government-sponsored) mechanisms which would reconcile the calculations by decision-makers of their private and organizational costs and benefits with total (economy-or sector-wide) costs and benefits. Perhaps the most widespread sort of nonmarket failure is the divergence between the officially set objectives of a public intervention program and the internal goals of the bureau or of the agency which is designated to carry out the program. Public objectives will typically suffer from ambiguity in definition and difficulty in measurement. In the absence of good direct-performance indicators, the bureau will set its internal objectives to guide and evaluate its own performance and the performance of its personnel. The consequent structure of rewards and penalties results in an internal version of the price system. For a concrete illustration the reader is invited back to Box 3.

To the extent that the maximization of internal goals is not synonymous with the maximization of public objectives, internalities will be present and will as predictably skew the results of nonmarket activities away from a social optimum as externalities may distort output in private markets. As Wolf puts it, the existence of externalities means that some social costs and benefits are not included in the calculus of private decision-makers. The existence of internalities, on the other hand, results in "private" or organizational costs and benefits being included in the calculus of social decision-makers.

A Last Word on Industrial Policy

As Watson has remarked, it is one thing to recommend wise government, another thing entirely to say how to make it Watson, op. cit., p. 207. operational. The main theme running through this chapter is that while the logical arguments underlying governmental intervention in favour of innovativeness are quite plausible, their empirical foundations are weaker. The true difficulty comes with the implementation of industrial and strategic trade policies: market failure may be converted into non-market failure.

In the final analysis the debates preceding and the actual setting in place of such policies in Canada so far represent an uneven contest. The proposers and implementers of such policies are full-time, highly paid public servants, backed up by consultants and all manner of techno-science lobbies, whose almost sole task and raison d'être is to dream up ways of spending the taxpayer's money. The resisters appear to be a few editorialists, a handful of academics, some think tanks and, occasionally, a business group, such as the Canadian Chamber of Commerce. As is the case with most policies that are of benefit to narrow sectoral interests, the general public cannot be mobilized to defend itself, since the costs of such defence may exceed the worth of non-intervention in any particular case.

While political competition in the United States and in Britain has shifted the onus of proof on industrial policy advocates, in Canada the restraint comes not from a sound, empirically-based argumentation, but from general deficit-induced restrictions.


Chapter 3 Innovativeness-An Economic Perspective

"Perhaps the greatest obstacle to understanding the role of innovation in economic processes has been the lack of meaningful measures of innovative inputs and outputs." Acs and Zoltan J. Acs and David B. Audretsch, "Innovation in Large and Small Firms,"American Economic Review, September 1988, p. 678.Audretsch

Introduction With Definitions

THE FIRST TWO CHAPTERS PRESENTED an overview of the issues to be discussed in this book and their inherent interest. It is now time to be more specific. In order to understand the soundness or otherwise of public policy interventions aiming at innovative performance, we first have to gain a more thorough comprehension of the innovativeness phenomenon itself. Two viewpoints complement each other: the economic and the managerial. The first pays more attention to the context, the second to the inner workings of innovativeness. This chapter focuses on the economic perspective.

It is not usual to speak of innovativeness. We use this word here to encompass both the creation and the adoption/diffusion of innovation. As was already stated in chapter 1, innovation is a process which, through technical, industrial and commercial steps, leads either to the marketing of new and improved products or to the commercial use of new and improved production processes, or both. (The research behind such innovations is sometimes called product-oriented or process research).

A linear vision of the place of both innovation and adoption/diffusion is offered in Figure 7. Basic research (or fundamental research), applied research, and development all feed successively into innovation. They are defined as follows:

Basic research includes research projects which represent original investigation for the advancement of scientific knowledge and which do not have specific commercial objectives, although they may be in fields of present or potential interest to the reporting institution.

Applied research encompasses research projects which represent investigation directed to the discovery of new knowledge and which have specific commercial objectives with respect to either products or processes.

Development is a technical activity concerned with non-routine problems encountered in translating research findings or other general scientific knowledge into products or These are definitions adopted by the OECD Frascati agreements.processes.

Click here to view Figure 7: A Linear View of Innovation/ Adoption

There are two points to note in connection with the definitions. The first is the difference between basic and applied research, a distinction that hinges on the researchers' motivation: new knowledge for its own sake, or practical objectives. The second is that development does not include marketing research or other work concerned with market development.

An important implication of these definitions is the distinction between invention, research and development, and innovation. Invention, whether representing a major or a modest advance in industrially useful knowledge, is merely an idea or a model which must be further developed, most often with the assistance of research and development activity, to reach a stage of technical feasibility. Once there, products and processes earn their adjective "innovative" only when they have been launched, with or without final commercial success, into a firm's market or into the firm's own production system. Furthermore, innovation is deemed to generate revenues for its creator (product innovation) and decrease costs, or equivalently yield productivity gains, for the customer-adopters to whom it has diffused. While this representation of the innovative process may be pedagogically advantageous, it is certainly oversimplified.

Thus, Mowery and Rosenberg remind us that Sadi Carnot, a French engineer, while working on improving the efficiency of steam engines, created the science of David C. Mowery and Nathan Rosenberg, Technology and the Pursuit of Economic Growth, Cambridge: Cambridge University Press, 1989, pp. 12-13.thermodynamics. Half a century later, in 1870, Pasteur was trying to solve practical problems with fermentation and putrefaction in the French wine industry. Along the way he invented the modern science of bacteriology. In the 1930's, work at Bell Labs on static in radiotelephone transmission resulted in the birth of radioastronomy. In general in industries at the edge of technology the problems encountered may be so new as to require a general rethinking of existing principles, or basic scientific research. Often, then, empirical technological challenges of the commercial kind lead to fundamental reflection, or to fundamental research.

Mowery and Rosenberg also point out that the distinction between basic and applied research turning on motivation is not robust. Individual scientists in private industry may genuinely pursue the advancement of knowledge per se, while their managers may be strongly motivated by expectations of useful findings.

It is thus possible that there is a very blurred line between research results that are appropriable and those that are not, despite the general belief that fundamental research falls into the latter category. As pointed out in the preceding chapter, the notion of non-appropriability of research results may be further weakened by the realization that basic and applied research may be required to enable recipients of potential spillovers to absorb such.

While the distinctions between fundamental and applied research are not what they are cracked up to be, their basic employment and usefulness cannot yet be discarded.

The Economic Council of Canada and the federal Department of Industry, Trade and Commerce carried out a survey of profitable product and process innovations in five Canadian manufacturing industries in Dennis P. De Melto, Kathryn E. McMullen and Russel M. Wills, Preliminary Report: Innovation and Technological Change in Five Canadian Industries, Economic Council of Canada Discussion Paper No. 176, Ottawa: October 1980.1980. The information, gathered about some 230-odd innovations, provides us with reasonable estimates on the relative outlays necessary to launch a new product or process-as outlined in Table 5.

Click here to view Table 5: Expenditures on the Development of Innovations by Process Stage for 234 Innovations

The figures in Table 5 are largely in conformity with North American patterns of industrial-goods manufacturers. (For a detailed anatomy of innovation expenditures consult Edwin Mansfield et al., The Production and Application of New Industrial Technology, New York: W.W. Norton, 1977.Mansfield et al. ) Between 40 and 50 percent of the total launch cost is accounted for by R&D expenditures, with most of the remainder consisting of manufacturing start-up costs. (We would expect marketing start-up costs to be much higher in consumer-goods industries.) Manufacturing start-up-takes a larger percentage in process innovations-the result of research aimed largely at within-firm cost reduction-than in product innovations which stem mostly from sales revenue-oriented research.

As expected, basic research activities that could be assigned to the reported innovations represented but a small fraction (3 percent) of the total investment; even as a fraction of total R&D they amounted to only 7 percent. Industrial firms tend to spend relatively little of their research funds on basic research, proportionally more on applied research, and the most on development.

Five years later, Statistics Canada reported a more generous engagement in fundamental research, still in the private sector, but this time by industry (see Table 6). While the first table shows more money spent on process research, the second implies heavier expenditures on product research. (In dollar terms, not shown in Table 6, transportation equipment, electronics and chemicals dominate in R&D outlays.)

Click here to view Table 6: Current R&D Expenditures, by Application and by Industry, 1986

A third table (7) gives yet another aspect of the part that fundamental research plays in the overall research effort-this time of nations (if, of course, it can be taken for granted that government does finance most of the basic research going on).

Click here to view Table 7: Non-oriented Research Programs as a Percentage of Civil Government Outlays on R&D

There is good logic behind the fact that most of the basic R&D goes on in the non-profit sectors. In advanced industrial countries there are several publicly supported sets of institutions devoted to "pure" advancement of knowledge, foremost among them the universities and government-sponsored research bodies, such as the National Health Institutes in the U.S., the National Research Council in Canada or the Max Planck Institutes in Germany.

Working on projects with barely identifiable commercial pay-offs, they nevertheless make research results actively and willingly available to the industrial sector where "practical" pay-offs may be perceived. This pattern of publicly supported non-profit research organizations devoting themselves to fundamental research has evolved for three reasons. The oldest of those reasons is the emergence of the university as the spearhead of basic advances in human knowledge.

The second and third reasons overlap: The commercial payoff from fundamental research is so uncertain that a profit-oriented enterprise finds it difficult to justify investment in it. Because basic research is perceived as having high social returns (as opposed to private returns accruing to a firm only), society is willing to support it when private business is not. Yet this traditional explanation, one might almost say old-fashioned explanation, should be modified in light of the Mowery-Rosenberg findings discussed on a preceding page.

Despite its aversion to basic research with its vague pay-off possibilities, private industry incurs substantial risk in its innovative activities both with regard to applied and developmental research and with respect to launch outlays. Mansfield gathered data from 16 U.S. industrial firms, among whom there were four drug and two electronics producers, and derived from each firm certain average project completion probabilities based on a number of successful and unsuccessful Edwin Mansfield, op. cit.ventures. He made the distinction between technical completion (the R&D project achieves its technical objectives), commercialization (the new product or process is carried beyond test-market or pilot plant trial) and economic success (a rate of return on the project exceeding other non-R&D alternatives.)

Among the 16 firms, each of whom have carried out a large number of new-product or new-process projects, the average probability of technical completion was a reasonably "safe" 57 percent. Given technical completion, the probability of commercialization was 65 percent. (This means that 0.57 times 0.65, or only 37 percent, of projects launched were commercialized.) The probability of economic success, given commercialization, was reported (by 11 of the 16 firms) to be 74 percent. Thus, on the average among this sample of firms, only 27 percent (0.57 times 0.65 times 0.74) of projects undertaken in the lab returned an economic profit. In a previous investigation of 220 projects completed during the 1960s in a chemical and two pharmaceutical laboratories, the odds that a project would return an economic profit on its investment were 12 in 100, or one in Edwin Mansfield, Research and Innovation in the Modern Corporation, New York: W.W. Norton, 1971, ch. 3.eight.

In view of the inherent riskiness of innovative enterprise and of the putative high social-though not necessarily private-returns to it, substantial funding to industry and to non-profit oriented research is provided from the public While most studies of the failure of new products conclude that the launching of innovative products is risky, there appears to be no study comparing risks in other firm endeavours, such as the hiring and turnover of employees.purse. Box 4 gives both a numerical and a graphical expression to this statement. Note that the statistics do not include tax alleviations granted to the private sector, on the order of $800 million in 1989.

Click here to view Box 4: Expenditures on R&D, by Performing and Funding Sectors, 1991

Technological Innovation

Technological innovation, as we said, comes in two guises. It is either a change in the production function of the innovator or the offer of a new or improved product or service. The first, a process innovation, will decrease the costs of the innovating firm-and possibly enhance its revenues if it licenses the process to non-competing firms. The second, a product innovation, will enhance the innovator's revenues and induce changes in the production function (lower costs) of his customers.

The graphical representation of a process, cost-reducing, productivity-boosting innovation is in Figure 8 which has panels (A) and (B). To understand the figure we must first give some background information.

Click here to view Figure 8: Productivity-Increasing and Cost-Reducing Technological Innovation

The explanation of productivity improvement (cost reduction) rests on the notion of the production function. For a given state of technology, such a function shows the maximum rate of output of a product per period that can be attained from given amounts of inputs. Suppose that the product can be manufactured with the help of two inputs, labour and capital, in a process that can be more or less intensive in one or the other factor of production, depending on the factor's relative prices. In other words, capital (K) and labour (L) can be to some extent substituted for each other, if an output (Q) is to be produced in a given period.

Mathematically this can be written as:

(1)Q = A Ka Lb

with A as constant. (Note that this multiplicative, so-called Cobb-Douglas production function requires a positive level of each input to give any output). Figure 8, panel (A) shows curves, called isoquants (iso = same, quant = quantity), which are a locus of all combinations of capital and labour yielding a given level of output in a period.

Let us assume that the production unit's output is 100 units per month and that it is achieved by some combination of capital inputs (such as machine time) and labour inputs, determined by the isoquant curve 100s, where S signifies "start." Suppose now that the maintenance department discovers a novel way of reducing machine down-time and the same output can be achieved with a lower level of both inputs, though this does not affect the relative productivity of capital and labour. We can now write:

(2)Q = (A + c) KaLb

and represent this neutral technological change by the isoquant 100n. The isoquant is southwest of and parallel to 100s, an indication that the same output as before, 100 units, can now be produced with fewer units of input. This outcome conforms to the definition of increased productivity, which is either a higher output per unit of input, or an unchanged output for a smaller quantity of input. Here the relative productivities of K and L remain unaltered.

Suppose now that, stimulated by continuing wage demands, the production unit is able to purchase and integrate with the existing process additional equipment which increases considerably productive capacity without claiming supplementary staffing. This could be indicated as:

(3)Q = A K(a+c)Lb

and is represented by the isoquant 100k. Capital (equipment), more productive after innovation, results in labour-saving and, because more of K units are likely to be employed now, the innovation can be designated as capital-using (or labour-saving). A similar reasoning stands behind a labour productivity-increasing innovation (100L).

Since in the wake of the innovation process more units of the (existing) product can be manufactured, or fewer inputs for an existing level of output can be produced, the unit costs of production will of necessity decrease. This phenomenon and its consequences are depicted in Figure 8, panel (B).

Panel (B) indicates some of these phenomena. Before the innovation was in place, a unit of the innovator's product sold at a price of P1, which covered the long-run average costs (variable, fixed and normal profit), LRAC1, at the output Q1. The innovation lowers the costs of production as indicated by the downward shift to LRAC2. The innovator, instead of satisfying demand of Q1 units at price P1-and reaping profits indicated by the area P1R1D2C2-reduces price to P2. This induces customers to expand their purchases to Q2 units per period. The innovator is satisfied with profits P2R2E2C2. But to his private economic profit, indicated by the corresponding area, are joined benefits accruing to his customers. These are depicted by the area P1R1R2P2, an increase in what is called "consumer surplus": customers are now buying (Q2-Q1) more units per period and for each they pay (P1-P2) less.

We shall come back to panel (B) when discussing social returns. Now let us turn attention to the revenue-increasing innovation. The firm is at first a member of a perfectly competitive industry and earns no economic profit, as depicted in panel (A) of Figure 9. At the prevailing market price the firm just covers its marginal and average cost. We could imagine the product to be a run-of-the mill antibiotic.

Click here to view Figure 9: A Firm in Competitive Equilibrium and After Obtaining a Patent

Let the firm now offer a new antibiotic, protected by a strong patent (see panel B). For a number of years it will be a monopolist, a sole seller in the new drug market. It equates its marginal cost (MC) with marginal revenue (MR) (different now from demand, D, or average revenue, AR) and earns above-average (monopoly) profits, indicated by the shaded rectangle that delineates price less cost times units sold.

Those monopoly profits are presumably the stimulus for past and future innovative activities. On the other hand, the unit price is higher than it would be under competition (where price = marginal cost) and fewer units are as a consequence purchased by the market. This profit-price trade-off represents an eternal bone of contention between consumer advocates and would-be innovation stimulators. A big fight between the two camps erupted on the Canadian scene in the mid-80s, as will be discussed in chapter 6.

Click here to view Figure: R&D Spending Flows, 1991

The Adoption and Diffusion of Innovations

The speed with which new products or processes, embodied in innovative equipment or intermediate inputs, are adopted by customer industries or households has a major impact on productivity growth in the economy. Similarly, slow or fast diffusion of consumer product innovations clearly makes an important difference to the enhancement of consumption quality. Indeed, there is a Canadian viewpoint which holds that the characterization of Canada's technological performance as a failure in innovation might be as readily identified with failure in the adoption of Steven Globerman, "Canadian Science Policy and Technological Sovereignty," Canadian Public Policy, Winter 1978, pp. 34-45.technology.

The differences in the diffusion speed of an innovation are easily discernible in Figure 10, taken from a celebrated study of hybrid corn by Zvi Griliches, "Hybrid Corn: An Exploration in the Economics of Technological Change," Econometrica, October 1957.Griliches. It is evident that Iowa farmers were much quicker to adopt the superior corn than those in Alabama. Much of the difference between the rates of adoption among the farming areas could be explained statistically by demand and supply factors.

Click here to view Figure 10: The Diffusion of an Innovation: Percentage of all Corn Acreage Planted to Hybrid Seed

Thus the original starting date of the S-shaped curve for Wisconsin, about 1932, indicates that hybrid seed companies bred and offered the new strains to farmers in that state much earlier than in Texas (1941). The suppliers, who had to breed hybrids to area specifications, chose to start with the heart of the cornbelt first.

The speed of adoption and the proportion of total acreage planted correspond to the slope and the ceiling of the curve. Wisconsin farmers, planting a higher percentage of their land to corn faster than Texans, perceived a greater economic advantage to experimenting with hybrids, since even a minor improvement in yield resulted in substantial returns.

Mansfield, one of the keenest analysts of the diffusion process, proposes four principal factors that govern the speed with which the innovation's level of utilization approaches market saturation: (1) the extent to which the innovation offers an economic advantage over existing products or processes, (2) the extent of uncertainty surrounding early use of the innovation, (3) the rate of reduction of the initial uncertainty regarding the innovation's performance, and (4) the size of the investment required to try out the Edwin Mansfield, The Economics of Technological Change, New York: W.W. Norton, 1968.innovation. Two other factors, among others, are also noteworthy influences in the diffusion process: the extent of the seller-innovator's salesforce and advertising effort, and the degree of competitive pressures to which the customer industries are subjected.

We have good insights into the adoption and diffusion of innovative industrial processes, such as continuous annealing, high-speed bottling, and others, and some understanding of the rate of imitation by competitors of new industrial products. Given this knowledge it is possible to make rough guesses of the economic gains brought about by an accelerated rhythm of an innovation's adoption among industrial customers. However, to the extent that diffusion is, like the earlier stage of innovation creation, a learning process that takes place among users, it can be assisted by a more efficient provision of information. This may be a legitimate goal of government support by means of demonstration centres, early experimental purchases, and other means by which to play upon the four factors mentioned above.

As regards new consumer products, the adoption process and its determinants have been extensively investigated. What is perhaps less well known are the costs imposed by regulation which actually retard the diffusion process. The most glaring instance is that of new drugs widely used in the sophisticated economies of Western Europe whose tardy arrival-by five to ten years-on North American shores results in unnecessary suffering and economic Sam Peltzman, "The Diffusion of Pharmaceutical Information," in Robert B. Helms (ed.), Drug Development and Marketing, Washington, D.C.: American Enterprise Institute, 1975.loss.

Another point with regard to innovation diffusion merits attention in the Canadian context. It is generally acknowledged that one of the great potential benefits of the presence of multinational corporations is their technology transfer activity to host country Harry G. Johnson, Technology and Economic Interdependence, New York: St. Martin's Press, 1975, ch. 5.subsidiaries. Even though subsidiaries may be restricted in their export activities, they often are not charged a full fee for technology transfer from headquarters, enjoying easy and less costly access to the benefits of their parents' research. Therefore, it would seem that any policies directed against a foreign industrial presence will retard the innovation diffusion process.

Diffusion of technology also appears to play a very important part in the overall economic growth of countries. Intuitively we all feel that Japan's spectacular growth was due to its fast imitation (= adoption) of U.S. technology, at least in the fifth and six decade of this century. Economists studying growth speak of technological "catch-up," or of closing the technology "gap." The basic idea is that provided a country has some technological absorption capacity and is willing to foster it, the adoption of foreign technology will speed up productivity increases and so economic growth.

Fagerberg-see Box 5-constructed and estimated technological gap models using data from 1960 to 1983 from 25 industrial Jan Fagerberg, "A Technology Gap Approach to Why Growth Rates Differ," Research Policy, 16, 1987, 87-99. See also, for an indirect confirmation, John W. Kendrick, "International Differences in Productivity Advance," Managerial and Decision Economics, Special Issue, 1989, pp. 13-18.countries. His estimates indicate that the more distant a country is from the technological leaders, the faster it tends to catch up. As it catches up, its progress slows down: it becomes more costly to scale technological-and so presumably economic-heights. An indirect confirmation of this assertion is the influential article of Cohen and Levinthal, mentioned in chapter 2 in connection with Wesley M. Cohen and David A. Levinthal, "Innovation and Learning: The Two Faces of R&D," Economic Journal, 99, September 1989, pp. 569-96.spillovers. These authors stress-and confirm by their findings at the microeconomic, individual line-of-business level-that the costs of knowledge transfer can be substantial and in part determined by the ease with which learning may occur. Clearly, the more complex the technology is, the less easy it is to master.

Click here to view Box 5: A Technology Gap Model Estimated

Some Effects of Innovation and Its Adoption

Obviously the primary and most important effect is the innovation's impact on the firm's bottom line, as achieved either by cost decrease or revenue increase. Next come the effects of its adoption on customers or imitators, measured by their productivity change or by "social returns." Finally, the effects diffuse throughout the economy and contribute to the welfare of other nations as well.

Innovation Creation

Let us start rather briefly with the profit impact (more will be said on this subject in the next "managerial" chapter). Sometimes it is possible to evaluate profit impact directly, when the innovative product or process presents a breakthrough and grows fast to become a major part of the corporation's activities. Smith Kline and French's ulcer-curing drug "Tagamet," and Michelin's radial tire are examples.

In most instances, however, the improvements are of a minor, steady nature and profit changes are difficult to trace back to them. Perhaps the most obvious way to circumvent such obstacles is to take the firm's R&D expenditures as a proxy for innovativeness, capitalize them and then compare the rate of return (ROI) on this intangible asset to the ROI on conventional, tangible assets.

Given, however, that ROI does not tell us anything about the firm's R&D variability or the degree of risk that it is subjected to, a more comprehensive approach is to take account of the stock market's evaluation of a company's R&D success. Englander, Evenson and Hanazaki examined stock market data from a very large number of "high-tech" (research-intensive) firms from 1970 to 1986:

To the extent that new technology translates into higher profits for innovating firms, market analysts have an incentive to identify these firms, and the expected future profits will be discounted into the current price of the firms' shares, even though current sales and earnings are low. Hence, the price/earnings (p/e) ratios of high-tech firms may serve as an indicator of how much new technology the market expects these high-tech firms to produce in the A. Steven Englander, Robert Evenson and Masaharu Hanazaki, R&D, Innovation and the Total Factor Productivity Slowdown, Paris: OECD Economic Studies No. 11, Autumn 1988.future.

Over this period the p/e ratios varied from 1.1 to 1.5 times the overall market index. When New York Stock Exchange-listed high-tech firms were excluded, the p/e ratios of the more "junior" high-tech stock varied from 1.2 (1976) to 7.4 (1986) times the overall market index.

In Canada, Johnson and Pazderka estimated the impact of R&D intensity (R&D spending as a percentage of book value) on the (stock market-determined) market value of a sample of about 50 Lewis D. Johnson and Bohumir Pazderka, "Firm Value and Investment in R&D," in Managerial and Decision Economics, v. 14, 1993, pp. 15-24. The dependent variable is defined as (average) price per share times the number of shares outstanding.firms. Statistically significant results indicate a positive, fairly large influence exerted by research intensity on the valuation of the firms by the stock market investors over the period 1985-1988.

Can we therefore conclude that innovation-creation is invariably, or at least as a rule, a royal road to business success? Not entirely. There were no failed firms in the samples examined. And-as far as this writer can tell-there is no record of scholarly examination of the riskiness of R&D enterprise to the outside investor. More on all this in the next chapter.

Figure 8, panel (B) showed how a cost reduction consequent upon a process innovation benefits both the innovator-by enabling him to turn an economic profit-and the customers, by offering them a lower price on the existing product. We shall dwell on the reason why the innovator did not appropriate to himself as profit the area P1R1E2C2, but only P2R2E2C2 in a later chapter. Here it is merely to be pointed out (again) that customers received a "consumer surplus" corresponding to the area P1R1R2P2. Except for the last one of them, each one would have been willing to pay a higher price than P2 per unit; the addition of all the sales corresponding to the higher prices-up to P1-represents in value what is called consumer surplus.

Innovation Adoption/Diffusion

We have just pointed out the benefits of cost-reducing innovations to downstream customers. The bulk of economic improvements originating in technology stems undoubtedly from the purchase of innovative products. By definition innovative products, unlike the existing products sold at a lower price in the wake of cost-reducing innovations, require that the customer modify his existing production processes. This is why the word adoption, implying an active participation, is used.

The bank may need to train its staff to use a new software program thereby yielding a richer customer-base information bank. The farmer may follow new planting instructions for the better seed. The consumer must learn how to operate his microwave oven or compact disk player. In all instances the factor input mix in the production process is modified and in all instances, when the innovative product performs as expected, overall productivity of the adoption unit increases. This process is the same as that described in panel (A) of Figure 8.

The productivity increase in customers' operations is well documented for the commercial firm, hardly at all for the private household. The leading savant here is Nestor E. Terleckyj, "Direct and Indirect Effects of Industrial Research and Development on the Productivity Growth of Industries," in John W. Kendrick and Beatrice N. Vaccara, New Developments in Productivity Measurement and Analysis, Chicago: University of Chicago Press, 1980.Terleckyj. He examined, in a production function context, the influence of own-performed as well as supplier-performed innovative activity. Some of his results were already presented in Box 2. Here a more detailed explanation is offered.

More precisely, Terleckyj analyzed data from 20 U.S. manufacturing industries. An estimate was made of the (regression) relationship between the average rate of growth in productivity during the years 1948-1966 and 1958 R&D intensity (research cost to value-added ratio) in a highly ingenious manner.

Terleckyj used as R&D inputs expenditure (ratios) incurred both within the industry, and by that industry's suppliers in proportion to the share of the vendor industry sales going to the buyer industry. In this way the frequent and valid observation that an industry's productivity depends on technology embodied in purchased capital equipment and supplies in addition to its own technology upgrading efforts was accommodated. In a second round of estimates Terleckyj divided "direct" and "indirect" research expenditures into government-funded and industry-funded categories. His results indicated that returns to private direct R&D were about 30 percent in terms of productivity growth in these twenty industries. Indirect returns, that is returns to "purchased" R&D, were almost three times as high. No statistical indication of significant returns to government-funded direct or indirect research was found.

It is likely that these results on the whole underestimate the effect of R&D since, as repeatedly stressed, the production function approach neglects product quality changes. It is also possible that returns to government funding were underestimated if government support stimulated private R&D expenditures that were complementary to government-sponsored research. It remains that R&D's impact on productivity in U.S. manufacturing was impressive.

Economic Council of Canada researchers, Postner and Wesa, inspired by Terleckyj's methodology, tried to throw light on the relationship between the growth rates of productivity and industrial R&D in H. Postner and L. Wesa, Canadian Productivity Growth: An Alternative (Input-Output) Analysis, Economic Council of Canada, Ottawa, November 1983.Canada. (Please see Box 2). The study is so complex and extensive that its summary is taken verbatim from the Economic Council's umbrella consensus paper The Bottom Line:

...was done for 13 manufacturing industries over the period 1966 to 1976, with both internal and contracted-out R&D being measured for each industry. The input-output method of measuring productivity growth was used in combination with the technique of multiple-regression analysis. The latter enabled us to distinguish the effects of this industrial R&D from the other variables known to influence productivity growth-i.e., changes in the amount of capital equipment of a given type and changes in the scale of operation. The results showed that R&D done in any given industry had little effect on its productivity growth but had a favourable influence on productivity growth in the industries that are supplied directly or indirectly by that industry. This means, essentially, that most of the industrial R&D in Canada that has a productivity-raising impact is oriented towards creating improved equipment and products for sale to other firms, rather than the development of new production processes that will be used internally by the industry performing the Economic Council of Canada, The Bottom Line: Technology, Trade and Income Growth, Ottawa, 1983, 27.R&D.

Innovation and Trade

One does not need to invoke neofactor proportions or neotechnology theories of international trade to sense that innovative products or lower-cost existing products give firms and industries an advantage in foreign markets.

The connection between trade balance and innovativeness can be represented in several guises. The proxy for innovativeness can be, for instance, R&D expenditure (an input measure) and productivity (an output measure). A vigorous investigation of the influence of R&D effort in a multifactor context on the foreign trade of 13 Canadian manufacturing industries was conducted by Petr Hanel, The Relationship Existing Between the R&D Activity of Canadian Manufacturing Industries and Their Performance in the International Market, Research Report, Dept. of Industry, Trade and Commerce, Ottawa: August 1976. The industries: food and beverage, textiles, clothing, wood products, paper, petroleum products, chemicals, rubber, non-metallic minerals, primary metals, metal products, machinery, electrical equipment, transport equipment except aircraft.Hanel. One of the most enlightening aspects of his study concerns Canadian export performance in its largest market, the United States. Hanel used as an index of that performance the share of United States imports in each of the 13 product groupings that Canadian manufacturing held against seven major industrial rivals, namely Japan, Belgium, Germany, France, Italy, the U.K. and Sweden, in 1969. He then related Canada's shares of the U.S. import markets to four measures of international competitiveness suggested by trade theory, read as of 1976:

Click here to view Table

Given that over 90 percent of the fluctuations in U.S. import shares was "explained" (R2 = .92) and that the effect of the R&D factor was highly significant, we can be reasonably satisfied that there is an association between R&D intensity and trade performance.

Interestingly labour productivity, a partial stand-in for "received" technological innovation, does not show up as statistically significant. Association does not, however, necessarily indicate causation. It is perhaps more prudent to envisage a scenario in which an original human or resource endowment gave rise to an industry successful in exporting part of its output and keeping its internationally competitive edge by investing in research leading to innovation.

While innovativeness in the above example was approximated by its input determinant, R&D, Soete used patents-an output approximation of innovation-to explain international trade Luc Soete, "The Impact of Technological Innovation on International Trade Patterns: The Evidence Reconsidered," Research Policy, 16, 1987, pp. 101-130.patterns. For each of the 40 manufacturing industries in 22 OECD countries he regressed the share of each country i exports of industry j in total OECD exports of industry j on the share of each country's i 1963-77 U.S.-registered patents in industry j in total OECD 1963-77 U.S.-registered patents in industry j; and on other variables. In other words, for a given industry j in 22 countries i

Export share ij = f (Patent share ij, Other variables)

In 28 industries out of the 40, the coefficient of patent share proved to be statistically significant and of the expected sign, thus confirming the notion that technological innovation as expressed in patenting confers an internationally competitive advantage. Contrast this, however, with our discussion of the trade performance of hi-tech industries further on in this chapter.

In concluding this section we remind ourselves that both R&D and patents taken out pertain to the "originating" industry and are far from measuring technology's impact in total. We give no estimate here of the adoption or diffusion impact on international trade performance among "receiving" domestic industries. (Though, as indicated by the example in Box 5, there are studies of the impact on industries abroad in a different context).

This brings us to consider again, as already done in part in chapter 1, important measurement issues connected with innovation.

Measurement Issues

We have bumped into several problems already with the measurement-and therefore with appraisal-difficulties with regard to innovativeness. Obviously the most serious one was the fact that innovation, and a fortiori innovativeness, can in most instances only be specified indirectly, such as by input (R&D expenditure or intensity) or by outputs, such as patents, trade balance etc.

Perhaps the next most difficult aspect is in the evaluation of the effects of innovation if these effects are deemed to be productivity improvements. We have presented, in Table 1, an estimate of the large contribution of technological change to the growth of the national income of the United States. Zvi Griliches, a leading U.S. econometrician and expert in the innovation field believes nevertheless that studies of productivity change seriously underestimate the contribution of innovativeness to economic growth, and that for two Zvi Griliches, "Productivity Puzzles and R&D: Another Nonexplanation," Journal of Economic Perspectives, 2, No. 4, Fall 1988, pp. 9-21.reasons. One is the difficulty of estimating correctly the magnitude of spillovers from one firm, industry, or country to another, a topic already touched upon in chapter 2. The other source of difficulty lies in the inability of researchers to measure productivity itself correctly:

The national income accounts, as currently constructed, do not reflect major components of the "product" of R&D and science and hence cannot serve as adequate measures of it. Most of the industrial R&D in the United States has been and is being spent on defense and space exploration connected projects. Its product is "sold" to the public sector and is measured, by accounting convention, by its costs. This implies a zero contribution to measured productivity growth except, perhaps, for its spillover effects on other products and industries.

The third aspect of measurement difficulty is also in part a reflection of definitional problems. Here we refer to the widely bandied-around and perniciously flexible expression "high technology" or simply "high-tech" firms or industries. At first, it would seem that the term is self-defined, that "high tech" designates an entity that relies for its economic success on technology. And this we believe would be the correct view. For instance, nuclear generation of electricity, engineering consulting, stockbrokerages and market research firms all rely on the latest in computer and telecommunication technology to a crucial degree. However, they are rarely, if ever, classified as "high-tech."

Hi-Tech

The last three sectors mentioned share the opprobrium of being classified as services and the last two of not undertaking much R&D as officially defined. It so happens that politicians still suffer from a manufacturing fetish. The result is a wealth of information on that sector, which accounts for about 20 percent of the gross domestic product, and very little for services that make up at least three times this Herbert Grubel and Michael Walker, Service Industry Growth, Vancouver: The Fraser Institute, 1989.proportion.

And so there is little interest in exploring hi-tech aspects in the service sector. More importantly, however, the sobriquet "hi-tech" is usually given only to industries whose firms are "own-performed"-research intensive.

Consider the scheme below which shows the possible sources of new technology for a given firm or industry:

•R&D performed intramurally or funded by industry

•"Invisible R&D imports" from affiliates abroad

•Technology-related payments (royalties etc.) to non-residents

•Government intramural R&D assignable to industry

•University research assignable to industry

•Commercial suppliers' R&D (first upstream level)

R&D performed by the firm or industry, as a percentage of sales or value added, or R&D personnel as a percentage of employees, is the standard yardstick by which industries are classified into high, medium, and low-tech. But there is a proliferation of these yardsticks, as indicated by the 8 types of classification listed in Table 8. All of them are or have been employed.

Click here to view Table 8: Methods of Classification of High Technology Products used in the United States and Canada and by the OECD According to Conklin and St. Hilaire

An essential point emerges from this rich confusion of definitions for the Canadian case: the different definitions-and so industry classifications-yield substantially different estimates of export-import trade balances of high-technology commodities. This is an indicator considered to be very significant by Canadian industrial policy advocates. (Recall Figure 6).

Now consider the next line in the enumeration. "Invisible" R&D imports are transfers of technology between multinational affiliates that are not formally billed to the receiving units. As will be shown in another chapter, they represent a crucial input into the innovativeness of the heavily foreign-owned Canadian industry. Neither this item, nor all of the following ones are counted in the R&D intensity measures used to define "tech" categories.

Technology-related payments do not necessarily signal technological underdevelopment. At least until recently, the Japanese had a large negative trade balance here, while the British had a positive one.

It is suitable now to return to the role of hi-tech industrial groups in foreign trade. Figure 6 showed the negative trade balance Canada has in hi-tech products. But hi-tech products do not an entire merchandise trade balance make. This was shown in a most convincing manner by Hughes, who analyzed the trade patterns of six countries in the Kirsty S. Hughes, "Technology and International Competitiveness," International Review of Applied Economics, 1992, 166-183.1980s. The three countries with the best overall trade performance during this time-Japan, Germany and Italy-had their best performances respectively in high, medium and low technology. The two countries with the worst performance-the U.S.A. and the U.K.-had their best performance in high technology. Hughes concludes that trade performance cannot be explained by reference to specialization by technology group alone.

Elsewhere we documented the importance of research carried out by (federal) government labs and by universities which directly benefit industry, and attempted with some success to assign it to specific Kristian S. Palda, "Technological Intensity: Concept and Measurement," Research Policy, 15, No. 4, August 1986, pp. 187-198.sectors.

We have already shown the role suppliers play in selling innovations downstream. We list a separate line for the first upstream level to signal that a powerful connection exists between such suppliers and customers. It is now well known that immediate customers often have a very active hand in the supplier's development of new materials, processes or capital Eric von Hippel, "Successful Industrial Products from Customer Ideas," Journal of Marketing, January 1978, pp. 39-49.equipment.

Thus, we conclude that the employment of the "hi-tech" concept is non-operational and perhaps even self-serving when used as a lever to obtain government favours, unless it is the case that "hi-tech" signals so-called strategic industries. Conklin and St. Hilaire cite an OECD 1985 paper that lists the characteristics associated with hi-tech David W. Conklin and France St. Hilaire, Canadian High-Tech in a New World Economy, Halifax: Institute for Research on Public Policy, 1988, pp. 136-7.products:

•high dependence on a strong technology base and a vigorous R&D effort

•considerable strategic significance to governments

•long lead-times from basic research to industrial application, short lead-times in commercialization, and accelerated obsolescence under the competitive pressure of new product and process introductions

•high risks and large capital investments

•high degree of international cooperation and competition in R&D production and worldwide marketing

While these characteristics do not constitute tangible criteria, there is a wide consensus, at least among politicians, that some industries are more important than others, that they are the sources of technical progress for the economy, that they are in a sense "strategic." And here opinions have been translated into actual support programs in Canada, France, Japan, and elsewhere, as will be described further on. Some indication of it is contained in table 9.

Click here to view Table 9: Public Funding for R&D in the High, Medium and Low-intensity Industries as Respective Proportions of Total Public Funding of R&D, 1980

Richard Nelson is probably the best-known authority in innovation economics which examine the proposition that hi-tech industries are strategic and Richard R. Nelson, High Technology Policies: A Five-Nation Comparison, p. 1.leading. The proposition that hi-tech industries are leading means that they tend to drive and mold economic progress across a broad front. The idea that hi-tech industries are strategic implies that national economic progress and competitiveness are dependent upon national strength in these industries, and help from government is required to shore up this strength.

Nelson examined only three of the important, conventionally-defined hi-tech industries across the five largest industrial nations: semiconductors and computers, civil aircraft, and nuclear power. He came to the conclusion that the first two were definitely "leading," in that they had widespread economic ripple effects. His opinion on their "strategic" nature was sceptical in that he was uncertain that taxpayer support for them was needed or helpful.

Since in this policy-oriented monograph the question of how useful taxpayer support is to innovativeness is the pre-eminent one, we submit that the concept of high-technology industries is, on the whole, detrimental to policy decision-making.


Chapter 4 Innovativeness-A Managerial Perspective

"No government program can substitute for managers, owners, and employees who are not competitive on their own." Terence Globe and Mail, July 22, 1992, p. B2.Corcoran

IN THIS CHAPTER WE LOOK at the firm's management of the generation or adoption of innovative products and processes. Understanding some of the basic managerial issues in this area should sharpen our critical appreciation of the scope and limits of public strategies to stimulate innovation.

Management-The Bridge Between Technology and Markets

It can not be repeated often enough that innovation-oriented research in a firm, or even commissioned by a firm, must be linked consciously and deliberately to commercial considerations. This link is provided by management both at the top of the enterprise, and by those in charge of the various functional areas: marketing, production, R&D, human resources, finance, management information systems, etc.

The most recent and probably the best illustration of the performance-tested strong symbiosis between R&D and the other business functions is presented in Box 6. In addition, the diagram shown therein illustrates the mutual influence of "environmental" variables and R&D intensity, in part already signalled in Figure 1.

Click here to view Box 6: R&D in Business Strategy

Apart from the marketing effort-R&D intensity connection indicated in the box, the other highest correlation of R&D is found to be with investment intensity (average investment divided by net sales). This could indicate that R&D efforts are a component of a larger strategic concept in which these efforts are coordinated with other investments. As the authors mention, R&D-intensive businesses must invest in order to capitalize on the opportunities presented by their innovation.

This is clearly another indication of the bridging role that management must assume to make innovation creation or adoption pay off in the market. It is well known that the conversion of a successful R&D investment into a successful venture in the market normally requires a commitment of people and capital many times larger than is required to do the initial R&D. The investment perspective of R&D again allows us to underline the importance of the business climate which either welcomes it or discourages it. It also makes us aware that an investment in innovation must be judged against other investments:

An expenditure on R&D is an investment, comparable to any other business investment made with the expectation of improving existing products or of developing new products and processes which will increase future profits. The amount to be invested in R&D is determined by comparison with other investment opportunities available to the firm. The attractiveness of R&D is influenced by its characteristics as an investment in comparison to these other opportunities.

The basic characteristics of R&D investment are the risk, the front-end investment, the delayed return and the time value of money.

The most important characteristic is that the return from an R&D project often occurs many years after the initial investment. The risk in that investment is higher because there is uncertainty concerning the financial market and the competitive situation when the results of the R&D program mature. By comparison a production machine has a short commissioning interval, and the risk of significant change in the business climate during the procurement and installation time is Report of the Business Council on National Issues and The Canadian Manufacturers' Association Joint Committee on Industrial R&D in Canada, Toronto: October 1979.small.

So a new product or new process is the result of an investment decision on the part of management. It seems natural, therefore, to enquire about the elements of management's deliberations about the desirability of such an investment. The first-and obvious-question that management will ask is whether, given the presumably higher risk levels, industrial innovations are profitable.

The Profitability of Industrial Innovations

An innovation, it has been said, is a new or modified product or process cast upon the waters of competitive seas. A successful innovation is by definition profitable, as it increases revenues or lowers production costs, but not all or even a majority of innovations are successful and therefore profitable. An innovation typically requires outlays not only for research, but also for "downstream" activities, such as new plant and marketing launch costs. Because the outlays precede the cash flows occasioned by the new product or process by more than one accounting period, we talk about investment in research and other innovative endeavours rather than about (current) expenses. In this sense we may legitimately ask about the profitability of innovation: did the money spent on one or several "lumpy" new-product projects return as much or more than money invested by the firm in joint ventures, new plant, different data processing systems? Because it is very laborious to obtain statistical evidence about returns to individual innovation projects, other means are usually resorted to in order to uncover some "typical" rates of return, and those only to the more frequently accessible data about R&D outlays.

The true profitability of a firm to its shareholders is reflected in the appreciation of the value of stock and the dividends paid out over some considerably long period. In the absence of this ideal measure, profitability is often approximated by some indicator of long-run return on capital employed. Nevertheless, the use of conventional accounting figures to calculate long-run returns carries some well known risks. Apart from the perennial problem of replacement cost, exacerbated by high inflationary rates, the prudent accountant's (and tax collector's) custom of expensing in its entirety investments in intangible assets distorts the picture of profitability of those firms which do research and advertise. In other words, the accumulating-and depreciating-investment in technological knowledge that is industrial research, is not properly capitalized. Nor is the continuing effort of the advertiser to build up the intangible capital of goodwill in the mind of the consumer.

Grabowski and Mueller, integrating a long line of previous studies (including this writer's), have shown a convincing way of calculating long-run rates of return on investment in research and advertising and present evidence that R&D yields above-average Henry Grabowski and Dennis Mueller, "Industrial R&D, Intangible Capital Stocks, and Firm Profit Rates," Bell Journal of Economics, Autumn 1978, pp. 328-343.profits.

Instead of defining the rate of return (ROI) in the usual way, that is, as:

PRt = (Salest - Variable Costst - dk Kt - rt - at ) / Kt

where

PRt = "profitability" or ROI in period t,

Kt = total (tangible) assets in t,
dK = depreciation rate for tangible assets,
rt = R&D expenditures in t,
at = advertising expenditures in t,

Grabowski and Mueller defined profitability as

PRt = (Salest - Variable Costst - dKKt - dRRt - dAAt) / (Kt + Rt + At)

where dR and dA are the depreciation rates for research and advertising assets, R and A.

This meant that they had to calculate, for the 86 U.S. corporations in the nine-odd manufacturing industries which constituted their sample, both research and advertising capital and their respective rates of obsolescence or wear out-in short, depreciation. They did so by assuming constant proportional depreciation rates for intangible capital:

Rt = rt + (1 - dR) Rt-1 and At = at + (1 - dA)At-1

In this formulation the stock of intangible capital (research or advertising) in a given period t is a sum of the current period's outlay (rt or at) and of the previous period's capital (Rt - 1 or At - 1), suitably depreciated. Using as depreciation rates 10 percent for R&D and 30 percent for advertising capital stocks, Grabowski and Mueller found that the proportion of total assets accounted for by intangible assets reached as high as 30 percent among the seven pharmaceutical firms in the sample. Interestingly, the variance of profitability rates among the sample firms was reduced by one-half when intangible stocks were capitalized, indicating the degree to which improper accounting treatment of such capital may distort the true economic picture.

Having defined and calculated a more realistic measure of profitability, Grabowski and Mueller related it then to the accumulated intangible assets that the sample firms held in research knowledge and advertising goodwill. In this way they obtained estimates of the long-run rates of return to these two types of investment. Holding some other potential determinants of profitability (market structure, firm size, industry growth) constant, they found an after-tax return on R&D capital of 11.7 percent. This is significantly larger than the average 7.1 percent average after-tax return on total capital employed by the firms in the sample. When the firms were divided into the 39 whose research-to-total-capital ratio was higher than 10 percent, and into others, the research-intensive firms (pharmaceuticals, chemicals, machinery) earned a surprisingly high 16.7 percent after-tax return on R&D spending, while the others (such as paper, metals and petroleum refining) did not obtain higher than average (i.e., 7.1 percent) after-tax returns from research. (Among the companies manufacturing largely industrial (i.e., non-consumer) products, advertising appeared to return above-average returns only in the case of pharmaceuticals.)

While the title of this section specifies innovation as the subject, so far only returns to R&D have been dealt with. The reason for this is an absence of statistical studies on innovation's profitability, due mainly to the fact that corporate accounting systems are not set up to report all the (R&D plus manufacturing set-up plus marketing expenditures, etc.) investments connected with putting an innovation on the market. Nor are they usually geared to report on the revenues connected with minor or even major product or process innovations.

Before providing some up-to-date information on R&D profitability, we had better explain the statistically established links between innovations and industrial research. In 1988 Acs and Audretsch examined 4,407 manufacturing innovations recorded by the U.S. Small Business Administration in The Small Business Administration defines innovation as "a process that begins with an invention, proceeds with the development of the invention, and results in the introduction of a new product, process or service to the market." Zoltan J. Acs and David B. Audretsch, "Innovation in Large and Small Firms: An Empirical Analysis," American Economic Review, 1988, pp. 675-689.1982. In this enormous sample they found a correlation of 0.746 = r between company-financed R&D expenditures in millions of dollars and the number of innovations launched by these individual firms. (The correlation decreased to 0.481 when company plus government-financed R&D was the measure). Next, assembling the data into 247 four-digit industries, A&A estimated the elasticity of (the number of) innovations with respect to (millions) of company-financed R&D to be about 0.4. These estimates indicate a reasonably strong substitutability of the R&D input measure for the "apples-and-oranges" measure of innovation.

Probably the best overview of recent R&D profitability studies can be found in the not easily accessible Economie et statistique journal in an article by Mairesse and Jacques Mairesse and Pierre Mohnen, "Recherche-Développement et productivité," Economie et statistique, No. 237-8, Nov-Dec. 1990, pp. 99-108.Mohnen. They discuss the findings of private rates of return to R&D in 5 studies of individual firms, in which the sample sizes range from 5,240 U.S. firms to 135 Japanese firms. The rates of return lie between 11 and 27 When industry dummy variables are not used in the regressions.percent. When industries rather than individual enterprises are the object of estimation, 4 studies (25 to 193 industries) show that private rates of return to stretch from 12 to 31 percent.

Bernstein, in two articles already discussed in Chapter 2 (see chapter 2, footnotes 17, 19) estimates the average private rate of return on R&D in 680 firms in 7 Canadian industries between 1978-81 to be 12 percent, while the average social rate of return (private plus spillover effects) is 22 percent; in 9 Canadian industries between 1963-83 these rates are estimated to be 32 and 58 percent.

Finally Zvi Griliches, "Productivity, R&D, and Basic Research at the Firm Level in the 1970's," American Economic Review, March 1986, pp. 141-154.Griliches, working on a monumental data set of firms ranging from 386 to 652 in size and covering the years 1967 to 1977, reached the opinion-similar to the earlier conclusions of Edwin Mansfield, "Basic Research and Productivity Increase in Manufacturing," American Economic Review, December 1980, pp. 863-873.Mansfield-that "firms that spend a larger fraction of their R&D on basic research are more productive, have a higher level of output relative to their other measured inputs, including R&D capital, and this effect has been relatively constant over time." Griliches finds, just as Mansfield did six years earlier in a smaller sample, that the private return on investment in basic research is several times higher than that on applied or developmental research.

The conclusion is probably warranted that private rates of return on R&D are higher than those on other private investments-if we are to believe a number of respectable studies. In Canada in particular this seems to be the case. We shall take up the subject of social returns to R&D when we discuss subsidies in a later chapter.

A mystery remains and casts doubt on the above conclusion. If ROI on R&D is so high, why has it not been driven down over time by profit-maximizing corporations who will try to equate returns to production factors on the margin? Such an occurrence has been found by Griliches with respect to the use of chemical fertilizers in U.S. Zvi Griliches, "Research Expenditures, Education and the Aggregate Agricultural Production Function," American Economic Review, December 1964, pp. 963-974. Yet the same Griliches remarks, 22 years later, again in an American Economic Review article, that one's response to this (mystery) depends on one's views as to the prevalence of equilibria in the economy. See Griliches 1976, op. cit.agriculture. And, of course, why do we not see money spent on R&D rise over time in Canada as a consequence of this profit-maximizing behaviour of private firms?

A tentative answer may be a high, but non-measured, level of risk connected with innovativeness. Profitability, after all, has two dimensions; the rate of return and its variability or risk. And yet, as we indicated in the previous chapter, both the New York and Toronto stock exchanges put a premium on the research intensity of member firms. Since stock market valuation does take account of risk, the tentative answer to the mystery of too high returns to industrial research remains so far unanswered.

Under certain circumstances, it is useful to enlarge the profitability perspective and examine other outcomes of innovation that are important to the firm. In a study of 43 European and 28 Japanese multinationals Kotabe found that a combination of market share, sales growth rate, and pretax profitability, called "performance" of a new product was strongly related to that product's innovation Masaaki Kotabe, "Corporate Product Policy and Innovative Behaviour of European and Japanese Multinationals: An Empirical Investigation," Journal of Marketing, April 1990, pp. 19-33."magnitude." In general, studies examining the success or otherwise of innovations, for reasons already mentioned, tend to concentrate on less "hard" measures than on profitability, as will be seen in the following discussion which turns to the question:

What makes innovations successful?

Having established that innovative endeavour can often lead to a profitable outcome, management is likely to ask next what particular paths to innovative products or processes could improve the chances of their success. This question can be put more simply as "what are the determinants of successful industrial innovations?" What comes immediately to mind is that such determinants can be classified as being either internal to the firm or largely influenced by the firm's environment. Management literature naturally concentrates mostly on the internal determinants, since these are the factors the firm can more easily manipulate. Economic investigations throw some indirect light on the role of environmental influences.

Understanding the importance of internal and environmental (or external) determinants of innovation is important. If there is a systematic pattern, a constellation of factors that add up to successful innovation, then this pattern could provide guidance both to management in its quest for increased profitability and to public authorities anxious to direct support to worthwhile projects with minimum risk. A quarter of a century of investigations has yielded some valuable results but their implications should be tempered by the realization that a panacea in this area is as difficult to come by as the stone of philosophers upon which, it is said, rests the success of the Japanese corporation.

Roughly speaking, there are basically two ways to examine the puzzle of innovation. The first zeros in on the innovative product or process, and considers the corporate management context as well. The second downplays the role of the individual innovation and concentrates on the firm that carries out innovative activities over some period of time. Both approaches try to single out the groups of factors that determine the commercial success of innovation. Perhaps the chief difference between the two is in their view of what constitutes commercial success. The innovation-centred approach tries to match the costs and benefits that are closely connected with the new product or process. The firm-centred or system-oriented approach considers innovative products or processes as just one of the constituent outcomes of a company's complex output, where the company's market success must be appraised in a long-run perspective.

The SAPPHO Project-Individual Innovations

The best known series of investigations concerning the managerial aspects of individual innovation was undertaken at Sussex University in the United Kingdom under the project name SAPPHO-Scientific Activity Predictor from Patterns with Heuristic Origins. The project was designed as a systematic attempt to discover differences between successful and unsuccessful product and process innovations in the chemical and scientific instrument industries of Western Europe and the United States over the post-war period to about Christopher Freeman, Economics of Industrial Innovation, London: Penguin, 1974, Chapter 5; Roy Rothwell et al., "SAPPHO Updated: Project SAPPHO Phase 2," Research Policy, November 1979, pp. 258-291.1970. The technique employed was one of paired comparisons, in which the two innovations, one successful and the other not, competed for the same market while not being technically identical. The criterion for success was commercial; a successful innovation was defined as one that obtained a worthwhile market share and/or profit.

Note that "worthwhile market share" and "worthwhile profit" are to some extent judgemental criteria; they have not been quantitatively defined in the study. The investigators themselves note that the overall success of an innovation must be measured by the total impact of the innovation on the innovating organization, and suggest that when share and profit figures give ambiguous results, "alignment with company strategy" could tip the scales of the success criterion. There remains the essential ambiguity inherent in accepting as success that which management defines as such; it is the non-removable bane of management studies that rely on interviews.

The comparisons were made against more than 120 criteria which at that time (1970) were believed to discriminate between success and failure in innovation. Some of these criteria are applicable to the innovation itself, some to the innovating firm. The flavour of the comparisons made in the 43 pairs of innovations (22 in chemicals, 21 in scientific instruments) is indicated by the partial results presented in table 10. Table 11 lists the individual innovation pairs.

Click here to view Table 10: Five Individual Variables that Discriminate Most Strongly Between Success and Failure of 43 Pairs of Chemical and Scientific Instruments Innovations

Click here to view Table 11: List of 43 Innovation Pairs

One-third of the 120-odd criterion variables proved to be statistically significant discriminators, at the 95 percent confidence level, between success and failure. These variables were also assembled into ten aggregate indices. Five of these together accounted for virtually all of the observed differences between success and failure. They are listed in table 12. The index variable "marketing," which correctly classifies 83.7 percent of the innovations, is a composite measure of the marketing effort deployed by the innovating organization and consists of the following individual variables:

•the innovation was part of a general marketing policy

•attention given to publicity and advertising steps taken to educate users

•sales effort a major factor

•a marketing decision (the observation of a need) rather than a production decision (addition to a product line created by new technology)

Click here to view Table 12: Five Index Variables that Discriminate Most Strongly between Success and Failure of 43 Innovations

In a similar vein, "R&D strength" is a measure of the performance of the development work on the innovation; "user needs" a measure of the efficiency with which market research or other procedures have established the precise requirements of the customer; "communication" a measure of the effectiveness of the innovating organization's communication network with the outside scientific and technical community. Finally, "management strength" is a measure of coordination. Four of the five index variables are represented by one of their individual "sub"-variables in Table 10, where the mnemonic abbreviations in brackets designate user needs, R&D, marketing and communications, respectively.

As can be expected, there are strong correlations between the scores of the five index factors. On average, the successful firm outperforms the unsuccessful firm in all five areas of competence for the given innovation. As the SAPPHO authors point out, one must look to multifactor explanations for success and failure in industrial innovation. Nevertheless, certain fairly simple conclusions do emerge from the study. Innovation is a coupling activity comparable to the blades of a pair of scissors. One blade represents the recognition of a potential market or in-house application for a new product or process. The other blade stands for unfolding technical knowledge, outside or inside the firm. The blades meet by matching the technical possibilities and the market needs.

The SAPPHO results unambiguously support the belief that firms that had a successful innovation paid more attention to the market than unsuccessful companies. The discriminating performance of the composite variables marketing and user needs attests to this. A strong in-house R&D (R&D strength) as well as good communications with outside technological developments in the general area relevant to the innovation proved to be another two of the five strongest discriminants between failure and success. As one of the SAPPHO project leaders remarks, the test of successful entrepreneurship and good management is the capacity to link together market and technology, by combining the two flows of information.

And, indeed, the fifth statistically significant discriminant composite index in Table 12 is designated management strength.

Screening for Successful Innovations

Building on SAPPHO and other studies Cooper took the next step and proposed a new-product model for predicting success or Robert G. Cooper, A Guide to the Evaluation of New Industrial Products for Development, Montreal: Industrial Innovation Centre, 1982.failure. The model was based on a survey sample of 186 industrial product innovations of Canadian manufacturing firms. Answers to the question about background characteristics of successful and failed innovations were factor-analyzed and aggregated into 13 larger dimensions. The degree of success/failure was then regressed onto these dimensions of determinants, of which 7 proved statistically significant. The model could predict success or failure in 85 percent of the cases.

An example is given to illustrate the procedure. Previously published research suggested that the innovation project's newness to the firm would detract from expected success. The survey questions (variables) touched therefore on whether the expected customers were new to the firm, whether the competitors to be encountered were new, whether the required production process was new, etc. Replies all loaded onto the factor "newness to the firm." The score of this factor was one of the regressors. It proved significantly negative.

The next step was to suggest a screening model called NEWPROD which is used for an overall rating on a proposed new industrial product project. Based on subjective opinions of within-firm or consulting evaluations or, in other words, on expert information at a stage at which no heavy project expenses are incurred yet, the screening model is proving highly successful and is periodically verified against new results.

Appendix 4A reproduces the first pages of the model's rating form to convey the flavour of this successful Canadian product.

We dwelt on Cooper's NEWPROD model for three reasons. First, reasonably successful attempts are being made to translate inquiries about the determinants of innovation into operational means to help with innovativeness. Second, the model shows that R&D is only one of the many elements entering management's forecast of a successful new product or process. Third, in the 48 tested questions submitted to project evaluators, there is not a single one that refers to the government's potential support, either in the form of tax relief or in the guise of subsidy.

This last point leads us to consider what external influences may influence the success of innovations. Clearly, market demand is the most powerful, but we chose to categorize it under "internal" since it is so close to the firm's concerns. Technological opportunity or changing technological environment is a powerful stimulant of industrial innovation and has documented influence on R&D expenditures, but does not necessarily contribute to innovative The Zif et al. article (op. cit.) offers evidence in this respect.success. Similarly, the competitive structure of the industry has been subjected to innumerable investigations with contradictory results, but again with R&D expenditure rather than innovative success as the variable to be accounted Cooper's count of "many vs. few competitors" is distant from the industrial organization inquiries; he finds that the more competitive the field is, the less chance of a success.for.

Finally, there is government support. Studies under this heading tend to be of a certain age, but they also tend to unanimity: government policies, whether by direct subsidy, purchasing schemes, merger encouragement or tax alleviations to encourage innovation have been, at best, One reference here is sufficient, more will be forthcoming further on. Rubinstein et al., "Management Perceptions of Government Incentives to Technological Innovation in England, France, West Germany and Japan," Research Policy, 1977, pp. 324-357.ineffective. Having examined at length aspects of individual innovations, we now turn to consider innovative firms.

The Louvain Studies-Innovation in Firms

In 1965 the Belgian Ministry of Science Policy asked a group of researchers at the Université Catholique de Louvain (UCL), led by Professor Ph. de Woot, to establish guidelines which would facilitate the allocation of public funds among enterprises asking for R&D support for specific Ph. de Woot and H. Heyvaert, "Management stratégique et performance économique," Economies et sociétés, 13, 1979, pp. 509-37.projects. The group undertook an extensive study of 96 firms representing 21 percent of Belgian firms engaged in industrial research and 69 percent of the total industrial R&D outlays. The principal hypotheses to be tested were that a firm's profitability increases with research intensity as well as with the size of the company, and that it is also influenced by the type of industry in which it participates. Profitability, however, as measured by a seven-year average rate of return on equity of the firm, did not prove to be statistically related to any of the hypothesized determinants.

What the researchers did find in the course of their survey interviews and data examination stretching over three years was, nevertheless, a definite, consistent pattern which it is instructive to compare with the SAPPHO findings:

1.While there is no discernible relationship between R&D outlays, or the number of innovations commercialized, and return on equity, R&D activity appears to be associated with higher profitability when

a)innovation project selection is a joint responsibility of top management, marketing, production, and R&D representatives;

b)a systematic market analysis is conducted in step with technical development;

c)explicit criteria are used to re-assess R&D projects with a view to abandonment.

2.Profitable firms adhere to a systematic product policy, practice a balanced allocation of human resources among production, marketing, and R&D functions, have a better knowledge of their environment, and have their executives spend more time in long-range planning.

These results formed the basis of another wave of research that formally tested the basic hypothesis that the economic success of a firm is largely determined by the quality of its management strategy, which in turn is reflected in four principal areas: a definite product policy; an equilibrium between the principal functional areas of marketing, production, research, top management; an internal and external information system; and a systematic innovation policy. A sample of 12 Belgian research-intensive firms of various sizes was chosen from fast-and-slow technologically changing industries for a thorough investigation. The basic hypothesis was handsomely confirmed.

Adoption of Innovations

Given that most of the firm's technology is "received" rather than developed in-house, economists and management scholars have devoted a lot of attention to this second aspect of innovativeness. Its importance is attested to by the already cited work of Terleckyj in the U.S. and Postner and Wesa in Canada (Box 2). A vivid illustration of the importance of adoption (and so, mutatis mutandis, of diffusion) of technological innovation in Canada is given by R.G. Lucas, "High vs. Low Technology: Assessing Innovation Efforts in Canadian Industry," Canadian Journal of Administrative Sciences, June 1986, pp. 121-145.Lucas.

Lucas looked at data over the 21 years from 1963 to 1983 emanating from two groups of Canadian industries:

Click here to view Table

Group A can be designated as low in R&D intensity, group B as high. Suppose now that most of the technological innovation in B is aimed at new or improved products, in A at more efficient processes. This seems likely if we consider the standardized outputs of group A. Both groups of industries depend on continued technological competitiveness, but group A buys most of it in the form of new process equipment from suppliers: it adopts the process innovations offered to it. (Undoubtedly, much collaboration takes place between suppliers and customers in the course of adoption and adaptation).

The statistical measure of process innovation Lucas used is machinery and equipment net fixed capital formation, MENFCF, which represents the acquisition of new machinery and equipment over and above that required for replacement purposes. It is an input approximation to innovation adoption, just as R&D is an input measure of innovation creation.

Table 13 indicates that while the median R&D intensity of group B industries is (R&D/Sales) = 5.8 percent, the same ratio for group A is only one-tenth of this. On the other hand the "low-tech" group's median process intensity, as measured by the ratio (MENFCF/sales) is 3 percent, or three times that of the "hi-tech" group's. The table tells a story not only of the other face of innovativeness, namely adoption, but also of the fragile distinctions between hi- and lo-tech designations.

Click here to view Table 13: Median R&D and Equipment Acquisition Sales Intensities Per Year, 1963 to 1983

Success in Adoptions

Adoption of an innovation by a buyer, whether of industrial or of the consumer variety, is the necessary consequence of its sale by the creator-innovator. In that sense, what makes adoptions happen is what makes innovations commercially successful. As discussed in the preceding chapter and in the preceding sections, the factors which make for speedier adoptions are manifold, but few appear to be government policy-influenced, unless, of course, one counts the framework policy stance of encouraging competition as a direct innovation policy.

Two aspects of innovation adoption, the other face of the coin of innovativeness, are worth mentioning as "determinants of success." Increasingly, management analysts observe and recommend greater cooperation between suppliers and customers (subcontractors and contractors, manufacturers and distributors, etc.). This applies a fortiori to new technology. In many instances it is nowadays difficult to make a distinction between innovators and adopters, both sides being responsible for the development of new products and processes. Box 7 reproduces an exhibit from a recent Canadian National Centre for Management Research and Development publication. Therein Professor More proposes an integrated model of developer-adopter cooperation for successful innovation.

Click here to view Box 7: Generic Decision Stages in the Development and Adoption Process

Let us mention briefly, as the other aspect, the success of governments in speeding up adoption, either of new technology emerging from its own labs or of private sector innovations. The latter endeavour finds expression in the funding and general support of innovation centres across Canada; the former in the transfer of technology from governments to the private sector, which has been one of the first mandates of the National Research Council. In part it is addressed by IRAP (Industrial Research Assistance Program), started in 1962. By fiscal year 1985/86 annual expenditures thereon reached about $40 million. From then on IRAP absorbed PILP, the Program for Industry/Laboratory Projects which started ten years earlier. PILP was even more focused on transfer, particularly to the small and medium enterprise. Together the two programs were spending over $50 million in fiscal 1987/88.

In his penetrating review of the effectiveness of certain federal government programs in support of technological innovation, Tarasofsky of the Economic Council of Canada stated that with respect to IRAP: "the information formally required from applicants is utterly incapable of permitting a rational judgment as to whether the project warrants Abraham Tarasofsky, The Subsidization of Innovation Projects of the Government of Canada, Ottawa: The Economic Council of Canada, 1984, p. 58.subsidization." He held a similar opinion of PILP.

A subsequent report by a governmental advisory committee found that "...a major shortcoming relates to the intramural capacity (of the federal government) for technology transfer. Although successive governments have made the commercialization of technology a high priority of science policy, the actual record is National Advisory Board on Science and Technology, Revitalizing Science and Technology in the Government of Canada, Ottawa, November 2, 1990. Released April 1991, p. 103.poor." It also cites with approval the United States Federal Technology Transfer Act of 1986 which provides incentives to federal government laboratories to transfer technology and whose consequences have been favourably reviewed in a 1989 Secretary of Commerce report.

Implications for Innovation Support Policies

The first purpose of this chapter was to emphasize that R&D is but one of the activities that management must undertake to innovate and to keep the firm competitive. As we have seen from the description of the SAPPHO and Louvain studies, the task of coordinating the innovative effort, of bridging the gap between the firm's technological supply and the market's demands is complex and requires high managerial skills.

Neither of these two studies or the normative NEWPROD model alluded to government support. However if there is to be direct support to private firms in the form of grants or studies, the policy implications here are that it should not be confined to the easily-defined target of R&D activity, nor should it go to firms with weak management. But the identification of good management in firms with little track record is difficult. Attractive projects, or even strategic plans for the innovation in question, require continuous monitoring, an activity difficult to achieve in public bureaucracies.

Beyond administrative difficulties there looms an issue which this writer, at least, has not been able to fully resolve: given that successful innovations are ultimately the result of strong management, public funds to stimulate innovation should go to well-managed firms. But well-managed enterprises are by definition innovative when market demands innovation, and do not need public funds to make them so. Should, therefore, public funds be expended on support to industrial innovation?

But let us not conclude on a "passive" note, and let us take up the point about subsidies to non-R&D innovative activities.

As we already mentioned, the risk of failure of an innovative product in its commercial stage is at least as high as the risk that the research and development effort devoted to the product will not reach successful completion. Similarly, the investment in the commercial side of the innovation, namely in manufacturing start-up and marketing start-up, exceeds by a large margin the investment in The De Melto et al. Economic Council of Canada study (footnote 4, chapter 3) covers only industrial products where marketing costs tend to be low. Even so, R&D constitutes only 42 percent of the total outlays.R&D. Thus the investment required to launch it and the accompanying risk are much greater in the typical innovative venture than the research outlays. This can be represented by a specific version of the product life cycle model, tailored to reflect the dangers facing innovative firms in industries with rapidly changing technologies.

In Figure 11 the development time, Dp, of the innovation is relatively long and costly, the introduction/growth time, I/G, is long, while the maturity period M is short as the market changes rapidly under the impulse of competitive innovations. (D stands for decline.) On a net present value basis, using the firm's capital cost as a discount rate, the investment project is probably barely, if at all, profitable.

Click here to view Figure 11: An Unprofitable Product Life Cycle

The point is that Canadian government granting agencies prefer to subsidize the R&D part of innovative projects, and that Canadian tax laws strictly exclude from its favourable treatment of research and development anything that goes beyond pilot plant outlays, and that more and more federal support to innovation is channelled through tax concessions rather than direct grants. The cumulative effect of these policies is that innovative enterprises are not supported in their most perilous periods. If subsidization of innovation is on the cards, then the whole innovation investment should be eligible for favourable treatment-not just R&D.

Click here to view Appendix 4A (Part 1): NewProd Screening Model: Rating Form

Click here to view Appendix 4A (Part 2): An Example of a New Product Project

Chapter 5 Le Mal Canadien?

"L'impression se répand qu'une malédiction particulière pèse sur la France... L'idée se crée, comme à d'autres périodes difficiles de notre histoire, qu'il existerait un mal français , une sorte de maladie chronique, insaisissable et spécifique. . . . Il n'y a pas de mal français."

Jean-Jacques From L'express editorial, August 23, 1976, cited in Alain Peyrefitte, Le mal français, Paris: Plon, 1976.Servan-Schreiber

Introduction

AS THIS CHAPTER WAS BEING WRITTEN, the annual World Competitiveness Report put Canada in fifth place among the world's 10 industrialized countries. Among the 330 criteria that enter the judgment some apparently touch on science and technology. "Canada ranks second in infrastructure, fifth in people and sixth in the extent to which government policies affect competitiveness. It shows poorly in science and Globe and Mail, June 20, 1991, B8. Since then Canada has been "demoted" to eleventh place. One of its worst scores is, however, still the one on technology, Globe and Mail, June 22, 1992, p. B2.technology."

What solid evidence do we have that Canada is doing poorly in technology-and so, presumably, in industrial innovation? Or, at least, what weak evidence and interpretation do we have? Now it may well be that on some absolute scale, present in the mind of sophisticated technocrats and gullible media, Canada is not doing well enough. Our measure of that, however, should be both time-series or cross-sectional numerical evidence. In less technical jargon, we should look at developments over time in innovation proxies for Canada, and at recent international comparisons between Canada and other countries. But because of what is obfuscatingly called "aggregation problems," we should also be hesitant about whole-economy, whole-country measures and prefer, where available, individual industry or sectoral appraisals.

Finally, it should not be forgotten that even if weaknesses in innovation or competitiveness are discovered, industrial policy measures to remedy such could prove either irrelevant, unpolitic, or inefficient. With this reminder, let us turn to the evidence on and opinions about le mal innovatif canadien. The narrative can be structured along the lines of Figure 1 in which input and output proxies of innovativeness are listed.

The R&D Input Proxy Economy-Wide Research Intensity

As stated, we look at the evidence both over time for Canada, and between Canada and other economies. The statistic most often employed here is the GERD/GDP, or the ratio of gross expenditures on R&D to gross domestic product. Table 14 gives the relevant figures. Over the eight years 1983 to 1990, total R&D expenditures in constant 1981 dollars grew at a respectable 4.1 percent per year. But that did not improve the 1.3 to 1.4 GERD/GDP ratio which is forever the target of complaints and testimony to failed government Over the last 15 years federal governments, both Liberal and Conservative, tried in vain to jawbone the private sector into helping to lift the ratio over 1.5 and up to 2.5.objectives. and in 1981 dollars (millions), 4,687, 5,110, 5,558, 5,841, 5,847, 6,000, 6,085, 6,231

Click here to view Table 14: Gross Domestic Expenditures on R&D (GERD) in Current and 1981 Dollars and in Percentages of the Gross Domestic Product, 1983-1990

This rough and ready ratio is most often used for international comparisons, as in Figure 3, wherein Canada is shown to be attaining less than half the ratios for Sweden, Germany, the U.S., and Japan. The patent absurdity of this comparison appears as soon as the composition of GERD is examined. Consider figure 12. There the black rectangles show the proportion of the GERD/GDP ratio that goes to defence R&D. For Canada, it is minuscule. In fiscal year 1986/87 it amounted to $221 million, or 8.6 percent of the federal budget devoted to R&D. These public R&D expenditures on defence approximated 3 percent of GERD in 1987. Compare this to about one-third of GERD devoted to defence in the U.S. and the substantial parts of GERD going to the same objective in France, the U.K. and Sweden.

Click here to view Figure 12: Composition of GERD/GDP, Selected Countries, 1987

Less visually easy, but more statistically convincing in this respect is Table 15 which lists 14 countries in order of increasing defence burden. Rank correlations are calculated between defence and overall research intensities, and confirm their positive links.

Click here to view Table 15: A Comparison of Defence and Economy-wide Research Intensities in Some OECD Countries, 1986

The picture is clear: countries that take upon themselves heavy defence responsibilities will necessarily incur heavy research expenditures. Such expenditures are a function of political commitments, not of innovative inclinations.

Sectoral Research Intensities-Macrosectors

The composition or the structure of an economy, as well as its defence responsibilities, matter in influencing research intensity. As of now the technological opportunity for investment in research is still the highest in the manufacturing sector, though service sectors such as software design are raising their research intensity. In the late 80s in Canada manufacturing R&D accounted for about 79 percent of BERD, the Business Enterprise Sector Research and Development expenditures.

However, the manufacturing sector in Canada is responsible for only 20 percent of GDP, while the 1987 or 1988 figures for Germany, Japan, and Sweden are on the order of 32, 29 and 31 percent respectively. It is thus natural that just on this account, Canada's economy-wide intensity is not fairly comparable to that of economies in which the manufacturing sector plays a larger role. A truer picture emerges in Table 16: Canada's weak showing in the GERD/GDP aggregate leagues improves somewhat after proper adjustment. A difference still remains, nevertheless.

Click here to view Table 16: International Comparison of Business Enterprise R&D Expenditures (BERD) as a Proportion of Gross Domestic Product: 1987

To see if this difference has narrowed over time, consider table 17 which lists the research intensity changes in the business enterprise sector for the G-7 countries. On this comparison Canada's BERD/GDP ratio has grown respectably both over time and vis-à-vis its big-time partners.

Click here to view Table 17: BERD/GDP Intensities and Growth, 1979 and 1990

Let us finally speculate, still within the section on macrosector comparisons, on what Canada's big-sector research spending would be like if Canadian research intensity were governed by the average OECD sectoral composition of GDP. The formula for the simulated expenditure is

Simulated $R&DSector,CDN = [($R&DSector,CDN / $GDPSector,CDN ) X ($GDPSector,OECD / $GDPTotal,OECD )] X $GDPTotal,CDN

The sectors that would seem "naturals" are agriculture, mining, manufacturing, and services. However, statistics on the first and the last sectors are not abundant for the major OECD countries. In a previous volume, figures from 1975 indicated that Canada spent well above the OECD average on research in agriculture and mining and about "par" in the service Kristian S. Palda, Industrial Innovation: Its Place in the Public Policy Agenda, The Fraser Institute, 1984, p. 86. The OECD countries: Austria, Belgium, Denmark, Finland, France, Germany, Italy, Japan, Netherlands, Norway, Sweden, UK, US.sector. The substantial shortfall was in manufacturing R&D.

Research Intensities in Individual Industries

Industry-specific reasons may account for a very large part of the difference between Canada's economy-wide research intensity and that of its industrially-advanced OECD partners. While we have already mentioned defence responsibilities and the overall economic structure of a country as important factors, on the individual industry level two important influences are prominent: the foreign-firm or multinational presence, and government intervention, either in the form of subsidy or of regulation. Only an industry-by-industry approach can draw out the particular circumstances determining industrial research intensity. This writer has been one of the first to advocate individual-industry comparative analysis across OECD economies and in 1992 the Science Council of Canada with Industry, Science and Technology Canada is publishing individual industry comparisons of a detailed nature.

For example, in the automotive sector, which is the largest apparent R&D under-spender, an explanation for the "shortfall" may be the integrated nature (read U.S. multinational presence) of the industry between Canada and the U.S. which, for Canada, results in low R&D-intensive assembly and higher R&D-intensive component ISTC, Science and Technology Economic Analysis Review, Ottawa: March 1990, p. 6.manufacture.

The analytical way to go about international industry comparisons of research intensities in individual three- or four-digit level industries was developed by this writer and a colleague in a background study for the 1983 Economic Council's K.S. Palda and B. Pazderka, Approaches to an International Comparison of Canada's R&D Expenditures, Economic Council of Canada, Ottawa: 1982 and Economic Council of Canada, The Bottom Line, Ottawa: 1983.The Bottom Line. The principal idea is that a particular, reasonably narrowly defined industry faces a similar set of environmental influences and responds to them in its pursuit of profit in a similar manner, no matter in which of the advanced industrial countries it is located. Thus, for instance, the decision to allocate investment funds to research and development by the Canadian pulp and paper industry will be governed largely by considerations similar to those relevant to Swedish paper firms.

If a model based on this reflection can account for a good part of variation in R&D outlays or R&D intensity (R&D/Sales) in one industry across a number of countries over a number of years, it will have captured the essence of the response of that industry's R&D spending decisions to the stimuli and impediments in its past and present environment. In order to evaluate the adequacy or sufficiency of the Canadian industry's research intensity, the following steps are undertaken:

1.For a given industry, such as pulp and paper, assemble data on R&D spending and its possible determinants for as many OECD countries and for as many years as possible; exclude Canada.

2.Estimate by multiple regression the relationship

[$RD/$Sales] = f[Determinant variables i,t] where country = i; year = t

The result is the average relationship between research intensity and its causal influences for a given industry in a number of OECD countries.

3.Insert the Canadian industry's values of the determining variables into the above equation and calculate the predicted value of Canadian research intensity for a given year.

4.Compare the predicted intensity with the realized research intensity of the Canadian industry in the given year. If the realized intensity falls short of the predicted one, there are grounds for supposing that this Canadian industry underinvests in R&D (and so in innovation?) when conformity to the OECD norm is assumed to be the criterion of optimality.

Table 18 shows the forecast and actual research intensities for two of the seven industries analyzed-paper and chemicals. As the table shows, the research intensity for paper was, on average, higher than the OECD's standard, while the chemical industry's intensity was lower. In this case the other OECD countries were France, Germany, Italy, Japan, Sweden, the United Kingdom and the United States.

Click here to view Table 18: Comparison of Forecast and Realized R&D Intensities ($RD/$Sales) in the Canadian Paper and Chemical Industries, (Percent)

The main advantage of this approach to the assessment of a Canadian industry's R&D spending adequacy is that it takes into account the variation of R&D determinants across the countries to be compared and, in a sense, holds it constant. Thus, for instance, an important influence on R&D spending, the degree of foreign ownership, is incorporated into the model and "neutralizes" the influence of invisible R&D transfers from multinational headquarters to subsidiaries. The main value of this model is diagnostic: if OECD norms are the relevant ones (and they are continuously being invoked in public debate), then this approach identifies industries that may be in need of R&D stimulus (or should not receive more of it). Furthermore, it points to those determinants which may be susceptible to government action.

This industry-by-industry international comparison can be envisaged as substantial progress toward shaping an innovation policy, provided that R&D intensity is accepted as the determining factor of innovativeness. It represents the first two steps in the following "global" approach to policy support for industrial Though not innovativeness, because considerations of adoption/diffusion are not present.innovation:

1.Define and measure statistically the gap between optimal and realized levels of innovative performance in Canada.

2.Estimate statistically the relationships between the gap and its several identifiable determinants.

3.Classify the determinants of the innovation gap as either unreachable or manipulable by policy-makers.

4.Consider pros and cons of the manipulable options.

5.Give policy recommendations, including the possibility of no policy.

Of the seven industries analyzed, three had higher R&D intensities than the regression had foretold, four had lower intensities than could have been expected given OECD norms. The estimates or forecasts just described represent both an over-time and across-countries perspective. Recently Hanel used an "as-if" way, similar in spirit to the formula given on page 121, to estimate a gap between the R&D investment performance of a group of seven Canadian manufacturing industries and that of their counterparts in 8 OECD Petr Hanel, "Ecart technologique de l'industrie canadienne," in Michel Leclerc (ed.), Les enjeux économiques et politiques de l'innovation, Quebec: Presses universitaires du Québec, 1990.countries.

Between 1979 and 1985 the growth in research expenditures of the Canadian group-electrical, chemical, transport equipment, metal fabricating, machinery, chemical-allied, others-was on the order of 172 percent over those 6 years. In the OECD group it was 131 percent. Nevertheless, the average total "shortfall" of the Canadian group remained at a substantial 51 percent.

What generally accounts for such a shortfall in individual (mostly) manufacturing Canadian industries? Let us look at the major reason, the presence of foreign subsidiaries in Canada.

Visible and invisible technology transfers

It is well documented that foreign-controlled firms in nearly all manufacturing sectors do a lot less R&D relative to sales than do domestically-controlled See, for instance, Economic Council of Canada, The Bottom Line, Ottawa: 1983, pp. 40-42 and Statistics Canada, Industrial Research and Development Statistics 1988, Ottawa: October 1990, p. 77. But the as yet unpublished results of a 1992 federal survey appear to contradict all previous statistics that indicated a lower research intensity among foreign firms.firms. The seeming paradox of foreign subsidiaries being heavily represented in R&D-intensive industries-intensive, at least, on the "own-performed" definition-and yet being less R&D intensive than their Canadian counterparts is resolved when it is recalled that such subsidiaries have less need to undertake research locally since they can rely on access to R&D results generated by their affiliates abroad. Statistically this explanation can be confirmed if either or both of the following conditions hold:

1.R&D- or technology-related payments abroad of foreign subsidiaries are relatively higher than those of Canadian-owned firms and most of these payments go to affiliated companies;

2.foreign subsidiaries benefit from invisible, that is free-of-charge, imports of R&D results from their parents or affiliates abroad.

The first condition is documented in Table 19. The table shows foreign- and Canadian-controlled shares of sales and R&D outlays before and after adjustment for R&D plus technology-related payments to non-residents: after adjustment the shares "even out."

Click here to view Table 19: Shares of sales, of R&D expenditures, and of R&D expenditures augmented (+) by technology-related payments to non-residents, 1975

Invisible transfer of technology

In order to verify that the second condition holds, namely that foreign subsidiaries benefit from invisible, free-of-charge imports, stronger assumptions must be made than with respect to R&D and technological payments.

The first assumption is that both parent/affiliate and subsidiary produce similar products using similar processes, and that R&D performed by the parent is applicable to, and tends to flow to the Canadian subsidiary. The second assumption is that the amount by which the subsidiary benefits from the whole group's R&D is proportional to its share of the whole group's sales. In other words, in order to remain competitive, the subsidiary requires a research intensity similar to that of the parent. This is a very conservative assumption since the fruits of R&D within a multinational enterprise are most likely to be of the nature of public goods: any member firm of the multinational group is allowed access to it, no matter what its "taxes," i.e. transfer payments to the group are.

This leads to the following formula for the invisible imports of R&D by a foreign subsidiary in Canada:

RD (Invisible) = RD (Potential) - RD (Canadian) - RD (Payments)

where

RD (Potential) = RD (Group)/Sales(Group) times Sales (Subsidiary)

RD (Canadian) = the subsidiary's intra-and extramural R&D

RD (Payments) = the subsidiary's payments to non-residents for technology aquired.

Note that the formula is conservative, not only on account of the public good aspect, but because technology-related payments (the only ones available at industry level) exceed R&D payments by about one third. Also, the formula does not preclude a negative figure: it is possible that a foreign subsidiary exports invisibly.

Because of the importance of this technology transfer, a detailed example of the calculation of 1981 invisible R&D imports is given here-for the three-digit SIC industry 374, pharmaceuticals and medicines.

The example shows a step-by-step procedure which is determined by the formula given on the preceding page. Potential R&D of foreign pharmaceutical subsidiaries in Canada is calculated by first compiling data on as large a sample of such firms as can be obtained (Table 20) and then "projecting" these data to a total population of foreign drug subsidiaries.

Click here to view Table 20: Global Sales and R&D and Canadian Sales and National R&D of 14 American Pharmaceutical Companies in 1981

Table 20 shows the group (world-wide) sales and R&D of 12 multinationals, and the sales and R&D expenditures of their Canadian subsidiaries. The potential R&D of these 14 subsidiaries is estimated at $35.6 million in 1981. To estimate the potential R&D of all foreign drug subsidiaries a "projection to universe" must be made:

a)calculate a multiplier

K = Sales of all foreign subsidiaries/sales of sample = 1,128 million/$623.2 million = 1.82.

b)assuming that R&D in universe is proportional to sales in the same way as in the sample,

total potential R&D = 1.82 times $35.6 = $64.8 million.

Next, the Canadian-performed R&D of all foreign drug subsidiaries must be estimated. Very likely a fairly accurate figure could be obtained from a special Statscan tabulation. In its absence the estimate of 1981 Canadian R&D is derived very simply. 1981 R&D intramural outlays in the pharmaceutical industry were $51 million. Using the 85 percent sales share held by foreign subsidiaries, we calculate (0.85 times 51) that about $43.4 million was performed by foreign subsidiaries.

Finally, net payments to non-residents for technology acquired (more than just R&D outlays) by the drug industry amounted to $7 million in 1981 (payments of $17 million, receipts of $10 million). Perhaps 85 percent, or $5.95 million, are the foreign subsidiaries' net remittances.

Using the overall formula, we can now estimate the total invisible R&D imports by foreign subsidiaries in the Canadian pharmaceutical industry to be:

= $64.8 - $43.4 - $5.9 = $15.5 million in 1981.

It is well known that the Canadian pharmaceutical industry, despite its high foreign ownership, is a heavy spender of company-financed R&D funds. Nevertheless, its invisible R&D imports amount to a quite considerable 30 percent of own-performed outlays. At the other end of the spectrum is the transportation equipment industry, SIC 32, which is largely foreign-owned and in which only the aircraft and parts industry, SIC 321, is domestically research-intensive. Estimated 1981 R&D invisible imports are $669 Calculations of imported invisible R&D for the transportation equipment and pulp and paper industries are given in Palda, Technological Intensity, Ottawa: Department of Regional Industrial Expansion, 1984.million. Somewhere in the middle lies the pulp and paper industry which is less than one-third foreign-owned and has a reasonable record of domestic research performance. Its invisible 1981 R&D imports are calculated to be $21 million. Thus for the 32 SIC transportation group, imported R&D represented about double ($669 million) the domestically performed R&D ($296 million); for pulp and paper, SIC 271, it represented only about 26 percent.

Using the approach determined by the formula, a Ministry of State for Science and Technology background paper estimated 1974, 1975, and 1976 invisible R&D imported to Canada. Table 21 shows that in 1976 these invisible imports amounted to $688 million, two-and-a half times higher than Canadian-located intra- and extramural R&D in these sectors. Thus, invisible R&D imports are likely to represent the most substantial component of all the R&D outlays that should be counted in to reach an adequate appraisal of an industry's technological intensity.

Click here to view Table 21: Invisible R&D to Canada, 1976

Statistics Canada applied the same methodology to later, 1983 data for transportation equipment (again) and chemical products industries. For transportation equipment the invisible imports amounted to a staggering $827 million, seven times the domestic R&D outlays. For chemicals it was $212 million, almost three times domestic outlays on research.

If we agree with this type of analysis and even if we scale down the estimates of invisible transfers by one half, we would find Canada's GERD/GDP ratio reaching 2 percent.

Now it may be said that importance must be attached primarily to industrial research carried out in this country for reasons of employment and training of scientists, for not missing out on fundamental trends in technological development, and even to defend national National Advisory Board on Science and Technology, Science and Technology, Innovation and National Prosperity, Ottawa: April 1991, p. 22.sovereignty. Such reasoning cannot be rejected out of hand, even if it is typically offered by self-interested industry, technology, and bureaucratic groups.

But there is an opposing viewpoint. Since the appearance of Adam Smith's The Wealth of Nations 200 years ago, the majority of economists believe that such wealth is best fostered when free market forces are allowed to seek out efficient, that is, lowest-cost solutions to economic challenges. The interest of the nation is taken to be synonymous with the interest of the buying public-the consumer in short. The consumer is best served when she has access to products or services benefiting from innovative modifications at the lowest price, no matter where such products or services spring from.

The consumer argument essentially states that it is immaterial whose lab the new product or process came from, as long as it is attractively priced. In the chain of cost components that add up to the final price, the R&D element is of some importance in innovative products. The lower it is, the less, obviously, the product's price to the buyer.

Government and university research on behalf of industry

Individual-industry research intensity is also typically understated by leaving out research tasks undertaken on behalf of industry, free of charge, in government and university laboratories. This writer has attempted to measure such costs Palda, "Technological Intensity: Concept and Measurement," Research Policy, 15 (1986), pp. 187-198.elsewhere. In a country in which the federal government performed $1.4 billion of research in its own facilities during 1989, accounting for 17 percent of GERD, some of this research must necessarily redound to the benefit of individual industries. On this it is useful to consult Box 4 which also shows that the federal government financed $629 million worth of university research (7.6 percent of GERD), with undoubtedly similar consequences. Unlike much government-performed research elsewhere, Canada's is mostly devoted to "civilian" technology, with a presumably closer connection to commercial outcomes.

But the government, usually so PR-conscious, is strangely silent on this issue. It does not in the least attempt to brag in comprehensive numbers about what approximate parts of the various ministries' budgets devoted to R&D seem to be close to the preoccupations of Canada's manufacturing and service This is not so in the case of the one-client ministry of agriculture.industries. The premier government research body, the National Research Council, overcautiously maintains that much of what it does cannot be assigned to individual SIC-type industries, since its research is often so advanced that the industries do not yet exist that may benefit from it.

It would require, therefore, a labour of Sisyphus to gather more than partial evidence on the pertinence of government-performed research to industrial interests. Table 22 contains a small illustration of the approach in which a tiny part of the annual disbursements of the Natural Sciences and Engineering Council to university researchers is assigned to individual industries.

Click here to view Table 22: Classification by 2-digit SIC industries of NSERC strategic grants to university researchers. December 1983, selected figures.

The conclusion drawn here is that an industry's alleged deficiency in technological intensity needs to be carefully examined before any taxpayers' funds are committed to support increased research effort.

The Trade Balance Output Proxy of Innovativeness

Rightly or wrongly, balance-of-trade statistics are often used as the principal criterion of a country's competitiveness on the world markets. We need not elaborate here on the cyclical and stage-of-development pitfalls of this measure. We must, however, register strong misgivings about deploring an individual industrial sector's trade deficit without considering the total trade balance. Furthermore, it is particularly pernicious to reason on share-of-world trade grounds amidst strongly expanding overall world trade.

Those who state that Canada is losing out in the innovativeness game, after showing us GERD/GDP ratios, point to the balance of trade in high-tech products. Consult Figure 6 to this effect. (The publication from which it is taken does not define hi-and the other techs, but it is likely to be OECD-2, as mentioned in Table 8.)

Table 23 covers the last 8 years of the high-tech trade shown in Figure 6. Exports, in current dollars, have more than doubled. The negative trade balance in 1981 constant dollars has not shown any tendency, and similarly the ratio of exports to imports and the deficit-to-GDP ratio. This table does two things that the doomsayers typically keep silent about: indexing to constant currency values and comparing to other important measures of the sector's or economy's performance.

Click here to view Table 23: Canada's Trade in High-Tech Products

Let us next turn to another trade indicator, to the balance of payments for technological services. Table 24 indicates, if anything, that by 1988 Canadian firms were not in deficit, and not very much prior to that as far as that trade is concerned. A more detailed examination by Hanel points out that telecommunications and office machines are chief contributors toward a positive Hanel, op. cit.balance. (Hanel also has astute observations regarding patents taken out in Canada by Canadians and foreigners as output indicators of innovativeness. We shall return to this topic shortly when we discuss diffusion).

Click here to view Table 24: Foreign Payments Made or Received for Technological Services, Selected Years

Finally, what about overall trade in manufactured products? Table 25 is one of the many possible indicators of trade performance, chosen because it does not rely on the share concept. Canada's performance over the years in manufactured exports does not seem to have been deficient compared to that of other industrialized countries.

Click here to view Table 25: Volume of Manufactured Goods Exports (1980=100) Selected Countries, Ranked by Increase, 1965-1989

Adoption/Diffusion as Part of Innovativeness

As we stressed earlier, the principal component of innovativeness is the firm's willingness and ability to adopt technology, whether from suppliers, licensers, or other providers. The mechanism of adoption/diffusion was explained in chapter 3 and its determinants briefly discussed; a lengthier enterprise-focused examination of adoption was offered at the end of chapter 4. But while we know quite a lot about individual adoption processes, we find it generally more difficult to speak of an economy's or industry's receptivity to technology than to comment on their creativity.

As usual, the extremes are easy to spot. All observers agree that Japan's progress to its present technological status was based on at least two decades of furious absorption of the foreign technical arts. Until recently, if not to this day, the technological balance of payments (transfer of patents, licensing agreements, provision of know-how etc.) of Japan has been negative, giving indirect evidence of technology OECD Science and Technology Indicators No. 2, Paris: 1986, pp. 55-57.absorption. The other extreme, of conservative ossification of industry thirty to fifty years behind modern practice, is the Soviet Union and its former vassals. Despite the best efforts of the KGB's industrial espionage directorate and the existence of certain exceptional branches of the defence industry, and despite a GERD/GDP ratio almost certainly quite superior to that of the U.S., the heavy hand of party control proved an insuperable barrier to fast adoption.

We have already mentioned patents: they are, if of domestic origin, indirect indicators of homegrown innovativeness; if of foreign origin, potential harbingers of technology transfer. The extreme tedium of patent statistics analysis which draws some conclusions on this score is illustrated rather well in an oft-cited piece published in the MacDonald royal commission N. Ellis and D. White, "Canadian Technological Output in a World Context" in D.G. McFetridge (ed.), Technological Change in Canadian Industry, Toronto: University of Toronto Press, 1985, pp. 43-76.series. The conclusions are highly interesting and encouraging:

Canada, it seems, fares reasonably well in world terms as a recipient of new technology. This country grants the third highest number of patents to foreign nationals, after the United States and Britain... Canada receives a large flow of detailed information on new technologies, which has the potential for use under licence by Canadian industry. The fact that much of this information is not readily available to other countries can only provide Canada with an advantage as a potential user... Canada, overall, responds well to changes in active and stagnant Ibid., pp. 72, 73, 74.technologies.

Against this optimistic conclusion there is a more pessimistic assessment by an earlier consensus report of the Economic Council of Economic Council of Canada, The Bottom Line: Technology, Trade and Income Growth, Ottawa: 1983, p. 63.Canada:

Our general finding is that new technology diffuses slowly into Canada from other countries. It also diffuses slowly from firm to firm and from region to region within the country.

A more variegated assessment is given by McFetridge and D.G. McFetridge and R.J. Corvari, "Technology Diffusion: A Survey of Canadian Evidence and Public Policy Issues," in McFetridge (ed.), op. cit., pp. 177-231.Corvari. They point out that diffusion and indigenous R&D are often complementary and that therefore firms undertaking R&D will be more receptive to the adoption of new technology: they will know where to look for it and know how to absorb it. Perhaps their most important observation is that the diffusion process is facilitated by the free movement of both goods and equity capital (for the latter, read multinationals) internationally. We have already dwelt at length on the invisible R&D imports which are the principal transfer of technology from abroad to Canada.

We recall here the evidence on purchases of machinery and equipment offered by Lucas (Table 13) and that gathered by Terleckyj on productivity increases among downstream purchasers (Box 2). This reminds us that an increased free flows of goods and services across borders, enhanced by the Free Trade Agreement with the United States (and soon perhaps with Mexico), will inevitably stimulate the rapid adoption of new technology by Canadian industry. What is perhaps more important is that it will offer all consumers a wider choice of technologically-efficient products and services.

The Competitiveness Issue

To be innovative, a firm, an industry, or an economy needs either to undertake some new product or new process creation, or to adopt rapidly relevant new technologies, or both. Innovativeness, vague as its definition is, undoubtedly feeds into the yet vaguer state of competitiveness. The mechanism behind this relies on enhanced productivity: new products and processes, invented or adopted, increase the efficiency (lower the costs) of productive processes in mining, agriculture, manufacturing, services, and in households. Thus, Michael Porter tells us that the only meaningful concept of competitiveness at the national level is Michael E. Porter, "The Competitive Advantage of Nations," Harvard Business Review, March-April 1990, 85; U.S Congressional Budget Office, Using Federal R&D to Promote Commercial Innovation, Washington: April 1988.productivity. He-and others-then go on to state that defining national competitiveness as achieving a trade surplus or balanced trade is inappropriate, for such do not guarantee a nation's standard of living. He concludes that to understand and explain competitiveness we must focus not on the economy as a whole but on specific industries and industry segments.

We have already looked at sectoral balance of trade figures and found them satisfactory within our context. It remains for us at least to glance at an economy-wide comparison, over time and between nations, of Canada's labour productivity in table 26.

Click here to view Table 26: Real Gross Domestic Product Per Employed Person: Comparative Levels in the United States and Eleven Other Countries

As of 1986, Canada's productivity level (GDP per employed person) was still ahead of the ten other nations with which we like to compare ourselves, and behind the U.S. Its productivity grew more slowly over the three-and-a-half decades than that of its ten competitors, but this is only to be expected if we believe in the catch-up hypothesis represented in Box 5. For a more thorough explanation see Moses Abramovitz, "Catching Up, Forging Ahead, and Falling Behind," Journal of Economic History, June 1986, pp. 385-406 and Richard R. Nelson, "Diffusion of Development: Post-World War II Convergence Among Advanced Industrial Nations," American Economic Review, May 1991, pp. 271-5.Note We can safely conclude on the basis of several Canadian studies-see the already cited works by Postner and Wesa as well as Bernstein, for instance-that Canada's overall productivity received a boost from new technology. Yet it would be hazardous to venture an opinion on how much of a boost and whether this was "enough."

Of course competitiveness does not live by innovation The glorious exception is the pharmaceutical industry where new products are the essence of continued survival. See Henry G. Grabowski, "An Analysis of U.S. International Competitiveness in Pharmaceuticals," Managerial and Decision Economics, Special Issue 1989, pp. 27-33.alone, although Porter, in his report on Canada's competitiveness, makes the point that "innovation-in its broadest sense-is the critical requirement for economic upgrading and increased Michael E. Porter, Canada at the Crossroads, Business Council on National Issues and Government of Canada, October 1991, p. 73.prosperity." And it could well be that it is precisely the governments who wring their hands over our alleged loss of competitiveness that are the institutions which are most responsible for it.

To cite the famous Porter Ibid.again:

What a new theory must explain is why a nation provides a favourable home base for companies that compete internationally. The home base is the nation in which the essential competitive advantages of the enterprise are created and The importance of the home base is discussed at length and doubted in Richard Lipsey's majestic draft of Economic Growth: Science and Technology and Institutional Change in a Global Economy, Toronto: Canadian Institute for Advanced Research, May 1991, pp. 168-175.sustained.

Do Canadian governments provide the right environment for such a home base? Here are three opinions and one newspaper story on this issue.

From a letter by J. Laurent Thibault, president of the Canadian Manufacturers' Association to Prime Minister Mulroney, dated September 25, 1990:

Excessive government demands on financial markets have also resulted in a run-up of Canada's foreign indebtedness to 35 percent of GDP or $230 billion in 1989. The resulting high interest rates needed to attract foreign money to finance the deficit means that Canadian industry is paying one of the highest real costs of capital in the world. There is also mounting evidence that the overall burden of taxes and levies of all kinds on industry is now too high and is hurting our competitiveness.

From Peter Cook's "Trade Failures at Home and Abroad" in the Globe and Mail, December 10, 1990, page B4:

Canada's high-interest-rate, high-dollar policies have led to one of the sharpest deteriorations in competitiveness and in export market share of any industrial country. . . . Measured in U.S. dollars, unit labor costs in manufacturing went up 18 per cent in two years, 1988 and 1989, while in the United States they went down 9 per cent.

From the Economic Council of Canada's Au Courant, No. 3, 1991 article on Canadian productivity.

The Council economists also believe that a significant change in the current Canadian fiscal and monetary policy mix would have a positive impact on productivity and competitiveness. Tighter fiscal policies (both federal and provincial), to address budget deficit problems, would simultaneously allow interest rates and the exchange rate to fall and thus improve Canada's cost competitive position. . . . According to their analysis, Canada's productivity and cost performance can also be improved by policies that: facilitate and strengthen the ability to adjust to longer-term structural changes by increasing market flexibility; and increase domestic competition by removing national and interprovincial trade and investment barriers.

Finally, from a story in the Financial Post, June 15, 1991, pages 1 to 2, "The Last Straw-NDP Final Blow As Ontario Faces Slow Flight of Investment Capital":

Three years ago, the signing of the Canada-U.S. free trade agreement brought John Wood's ambition of making his appliance manufacturing company a North American player closer than ever.

To serve that market, Wood, president of W.C. Wood Co. Ltd., had already bought 140 acres near Guelph, Ont. He'd always assumed the expansion would be there. He grew up in Guelph and the company his father started in 1930 employed all its 700 workers there, making freezers, humidifiers, and range hoods.

But a study comparing the cost of expanding in Guelph to Ohio gave him a shock. Serviced industrial land in Ohio cost about U.S. $7,500 an acre compared to $100,000 around Guelph. Municipal taxes were lower by 60 percent, corporate income taxes by 12 percent and interest rates by 33 percent. The study concluded the C$ would have to be at US65¢ to make a new plant competitive in Canada. Last summer, his new freezer plant employing 100 opened in Ottawa, Ohio.

"Ontario in particular and Canada as a whole are losing all of the benefits that they gained under the free trade agreement through items which we do have control over," says Wood, citing the high C$, interest rates and especially escalating government spending.

Conclusions

We have shown that many of the well-known complaints about Canada's lack of innovativeness do not hold up to careful scrutiny. If, therefore, it ain't broke, to cite the popular maxim, why call for fixing it? And we come again to the opinion that if policies are to be invoked to increase innovativeness and even competitiveness, they should be of the broad framework, target-neutral kind that would have kept Mr. Wood's new freezer plant in Canada. And, indeed, in the first annual report to the United States President and Congress, the Competitiveness Policy Council identifies as the first line priority issues for its attention saving and investment (read Government deficits and tax policy) and education and Building a Competitive America, Washington: March 1, 1992. Technology comes next, but the Council does not perceive the problems there as lying within the government's purview.training.


Chapter 6 Government and Innovation in Canada

"That is, when the tax system, subsidy programs and R&D contracts are taken into account and when R&D contracts are assumed to be 50 percent subsidy, Canada provides a greater incentive to engage in R&D than does any of the other 11 countries for which these kind of data were available."

McFetridge and Donald G. McFetridge and Jacek P. Warda, Canadian R&D Incentives: Their Adequacy and Impact, Toronto: Canadian Tax Foundation, 1983.Warda

Introduction

EVEN IF THE NEED HAD BEEN ESTABLISHED for an active industrial policy toward innovation in Canada-and the evidence of the preceding chapter certainly did not strongly point that way-another question should be raised before speaking to the kind of policy, if any, that would be appropriate. Has the support directed to the encouragement of innovation in Canada been clearly inadequate in level or ineffective in desired consequences? This chapter will attempt to cast light on the issue in a broad manner. We shall first discuss the scope and the kind of direct government support of innovation that has been available to industry in the past and is available now. This support is always undergoing change and no definitive listing of it can be offered. But it is, and has been considerable. We shall then present the normative case for the subsidization of innovative activities and the rules flowing therefrom, and ask whether there is any empirical evidence that such rules have been adhered to by the federal government. Next we will touch on some aspects of Canada's tax stimulants to R&D and discuss briefly international comparisons thereof. Finally, we shall look at some actual Canadian cases of government participation in innovative activities, from research in government labs to private sector subsidization.

The Extent of Past and Present Taxpayer Support-An Overview

The dollar amounts of federal and provincial expenditures-in house (intramural) as well as grants (direct subsidies) and contracts (partial subsidies of about 50 percent according to McFetridge and McFetridge and Warda, op. cit.Warda) have been outlined, for 1991, in Box 4. For ease of reference Table 27 reduces, simplifies, and adds 1984 statistics to the data provided in that box.

Click here to view Table 27: Funders and Performers in Canada's GERD, 1984 and 1991

Three sorts of information emerge from the table: first, Canada's federal and provincial governments finance-and perform-a substantial share of the country's R&D efforts; second, their overall share in GERD has decreased from 43 percent to 36 percent as funders and from 25 percent to 18 percent as performers; third, this decrease in GERD share has been picked up by the business sector. The second and third developments are certainly in line with OECD expert advice and opinions.

The size of the taxpayer's contribution to the nation's research undertaking is, comparatively speaking, on a par with Japan's and Italy's with respect to the fraction of GERD/GDP financed by public funds. (Just as the Japanese and Italians do, Canada's taxpayers finance only a negligible fraction of GERD/GDP's ratio devoted to defence). This, too, rates a favourable assessment.

So much for the alleged insufficiency of the Canadian government's-or better, the Canadian taxpayers'-contribution to innovation outlays, an insufficiency proclaimed almost weekly from opposition benches and by media pundits. Is the distribution of this largesse deficient? Box 4 indicates that the federal and provincial governments together funded $533 million worth of business enterprise R&D expenditures in 1991, or 10.2 percent of it. This is almost certainly quite a wrong amount, a substantial underestimate; it does not take into account the tax advantages granted to R&D performers.

We can surmise that about one-half of the $465 million of federal financing consisted of government research contracts to take care of federal departments' research in-house and policy decision-making needs. The other half consisted of outright grants to industry. Since it was estimated by McFetridge and Warda that about half of the value of contracts represents pure subsidization of industry, we end up with roughly three-quarters of $465 million as industry subsidies, or about $349 million in 1991. But to this must be added the tax credits available to the private sector, and, of course, the total deductibility of both current expenditures on R&D and capital expenditures on equipment (though not on land and The immediate deductibility of capital expenditure for equipment is not available in many other countries, such as the U.S.A.buildings).

Table 28 shows a rough estimate of the size of tax credits and reasonably accurate figures on grants. By 1988 the estimated tax credits represented two-and-a-half times the size of the grants, and the two together exceeded $1 billion in private sector subsidization, or 13.5 percent of GERD, or 22.6 percent of business enterprise expenditures (BERD). Note that no contracts are involved in this calculation-the subsidization is "pure." This brings us to consider, piecemeal, some of the granting and tax incentive programs, past and present.

Click here to view Table 28: Federal Grants Given and Investment Tax Credits Claimed with Respect to Industrial R&D

Direct Subsidy or Grant Programs

Table 29 reports on three taxpayer-financed subsidy programs for the ten years, 1980 to 1990. The oldest and most venerable of them (started in 1962) is run by the National Research Council. Called the Industrial Research Assistance Program (IRAP), it pays staff salary costs of selected research projects likely to initiate significant technological advance through commercial development and application in Canada. The selection criteria are the applicant's expertise in the field, and ability to effectively commercialize the research results. PILP, the Program for Industry/Laboratory Projects is designed to promote a more rapid transfer to industry of the results from NRC laboratories and other federal laboratories where there are important opportunities for Canadian industrial exploration. It has been administratively absorbed by IRAP since 1986.

Click here to view Table 29: The Three Major Federal R&D/Innovation Grant Programs, 1980 to 1990

The Enterprise Development Program, EDP, was created in 1977 and administered by the variously-named federal ministry of industry, nowadays and until further notice called Industry, Science, and Technology Canada. It was of assistance primarily to small and medium-sized businesses. One of its two goals was to support innovation; in that role it actually amalgamated and superseded a number of veteran granting programs run by Industry, Trade and Commerce. Grants were provided for new product or new process development projects which had the potential for profitable commercial exploitation. Up to 75 percent of the costs of approved projects could be contributed by the government to companies with sales of less than $10 million annually and up to 50 percent for larger companies. In 1980 the EDP's coverage was increased to provide special assistance to the electronics industry, and support for large-scale projects or company mergers that increased electronics production or research and development. Ottawa's Silicon Valley North was undoubtedly a prime beneficiary of these measures.

In 1982 the STEP (Support for Technologically Enhanced Productivity) program was introduced as a major extension of the special electronics fund (SEF) of the EDP. Given a budget of $20 million for 1982/83 it went beyond R&D to innovation adoption and diffusion support. At the same time, the scope of the Enterprise Development Program was enlarged to permit financing of market potential studies for innovative products and processes and other innovation-connected activities.

When the precursor to this volume came out, it carried on the above description by saying:

In attempting to list in a comprehensive and concise manner the most important innovation assistance programs of DRIE (formerly Industry, Trade and Commerce, and Regional Economic Expansion) one is overwhelmed by the dynamic inventiveness of Ottawa's bureaucracy in this field. The phase has apparently been reached in which there is such a plethora of these programs that a consolidation and simplification of them is required. According to Bill C-165 (passed in June, 1983) a new, all-embracing program designated as the Industrial and Regional Development Program (IRDP) is being put into place. Any company anywhere in Canada will be eligible to apply for aid, whether it seeks a grant to develop a new product, or to obtain a loan guarantee to restructure an outmoded manufacturing operation. Chances are that by the time this book reaches its readers yet another reorganization of DRIE innovation assistance schemes will be in the offing.

Well, the IRDP program, which was only partly designed to support innovation, was actually cancelled only about five years later and then slowly wound down. The confusion accompanying its early days was analyzed in Box 3. By the beginning of the 1990s we are back to a plethora of programs again, in part described in Table 3. The most complete and annually updated listing of these programs can be found in a CCH David Horsley et al., Industrial Assistance Programs in Canada, Toronto: CCH Canadian Limited, various years. See also the colourful brochures issued by ISTC, such as the Support for Technology and Development, Ottawa: 1989, which lists ten federal and at least thirty joint federal-provincial technology subsidization schemes. Finally, ISCT issues yearly reports in which the individual programs are described in detail.publication. If there are any direct successor programs to IRDP, they are to be found in the Atlantic Canada Opportunities Agency Action Programs and the Western Economic Diversification Programs.

The largest program in the quiver of the ministry (and the federal government) is the Defense Industry Productivity Program, DIPP, which has a substantial R&D/innovation component. It is aimed at defence industries and their subcontractors and designed to increase their technological competence for export activities. The aeronautics and electronics sectors are the most important recipients of grants and loans. Those grants and loans cover a substantial percentage of eligible projects. "Defence" has to be taken with a grain of salt. Aircraft engine development, traditionally a voracious consumer of these subsidies, is not exclusively of military interest. If there is any trace of strategic trade policy to be found in federal subsidization, it is perhaps within DIPP with its export orientation.

The preceding footnote mentions joint federal-provincial subsidy programs. But there are, as well, of course, purely provincial programs that dig into taxpayers' pockets to support exciting, if not necessarily economically viable innovation schemes. A splendid example is the Quebec Fonds de développement technologigue. Modelled in part after Ontario's Technology Fund, it was set up following the 1989-1990 budget speech and endowed with $350 million. Two of its important subsidy programs are les projets mobilisateurs (catalyst projects) and small and medium business R&D projects.

The catalyst projects are said to have stimulated research projects close to $200 million, at the cost of about $50 million of subsidy, by May Ministère des finances, 1991-1992 Budget, Québec: May 2, 1991, Appendix A, Section 2.3.1991. The grants are destined to precompetitive research: partnership between several enterprises and university research institutes or departments that aim to provide a technological solution which constitutes real progress and leads to an influential market position of strategic importance to Quebec's economy.

The generosity as well as the complexity of the scheme-when subsidy and tax savings are combined-is apparent from Table 30. The object of the illustration is a manufacturing corporation whose taxable income exceeds $200,000. The first horizontal part of the table looks at the cost of the expenses incurred outside of the enterprise which are eligible for the grant. These would be mostly expenditures for R&D incurred extramurally. The fund (FDT) finances 50 percent of the $5 million, or $2.5 million. Tax savings offered by Quebec (GVT means gouvernement) amount to $154,000, and by the federal government $646,000. The firm, therefore, pays only 34 percent of the $5 million cost, or $1.7 million.

Click here to view Table 30: Financement d'un Projet Mobilisateur

The second part of the table concerns (mostly intramural) R&D expenditure. Tax credits and deductions offered by both governments amount to $3,528,160, giving a net cost of in-house R&D to the enterprise of $1,471,840, or 29 percent. Thus the total cost of the putative $10 million project comes to 31.7 percent after subsidies.

In sum, these are impressive figures, even when compared to Canada's OECD, Industrial Policies in OECD Countries, Annual Review, Paris: 1990, part 4.competitors. Why is it, then, that the response of overall R&D spending is so anaemic, at least in the view of Canada's technology critics? A partial, perhaps even substantial answer may be the unimportance of industrial strategy when measured against the general receptivity-to-business climate. The question we just asked is, if anything, reinforced further by a brief appraisal of tax stimulants.

Tax Measures in Support of R&D

We have already indicated, with the help of Table 2, that Canada's tax régime is one of the most generous in the world with respect to R&D expenditures. We stress the R&D because we wish to make a distinction between industrial research, a component, and innovation outlays. Here we go into more detail, relying primarily on three sources of information: the already mentioned CCH Canadian compendium on industrial assistance programs in Canada, the Conference Board report on the international competitiveness of Canadian R&D tax incentives mentioned in Chapter 1, and a report prepared for Industry, Science, and Technology Deloitte & Touche, A Comparison of Tax Incentives for Performing Research and Development in Canada and the United States, Ottawa: May 1990.Canada.

For taxation purposes, the federal definition of R&D is accepted by the provinces. It states that "scientific research and experimental development is a systematic investigation or search carried out in a field of science or technology by means of experiment or analysis." It specifically excludes such activities as market research, sales promotions, research in the social sciences, routine data collection, and prospecting for minerals, oil and gas.

Federal tax support for R&D comes in two guises. The first are the deductions allowed. Firms can either write off immediately not only current expenditures on R&D incurred in or out of Canada, but also expenditures on R&D machinery and equipment, or they can choose to defer them to a future year. Ontario strengthens deductibility by offering a so-called super-allowance.

The second and, internationally speaking, exceptionally generous provision for support comes with tax credits. These can be applied by enterprises against their income tax payable and, nowadays in Canada for small businesses, can be obtained from the federal government as a cash refund if no tax is payable-such as when losses are incurred.

The tax credits are allowed on qualifying R&D expenditures, reduced by the amount of government grants and domestic contract payments, if any. For large corporations (roughly with taxable income over $200,000 a year) the federal tax credit is 20 percent of R&D expenditures incurred; 30 percent in the Maritimes and the Gaspé peninsula. The credit is increased to 35 percent for qualifying Canadian controlled private corporations (CCPCs), with an income of less than $200,000, up to $2 million of R&D expense-and 20 percent on any above that limit. The credit is, however, taxable as income the following year.

Further tax credits are obtainable under Nova Scotia and Quebec tax laws. For instance, in Quebec there is a 20 percent refundable tax credit available in respect of R&D wages, increased to 40 percent if the corporation is Canadian-controlled and has an equity of no more than $10 million and assets of less than $25 million. In addition, a 40 percent tax credit is available to corporations that contract for research with universities or engage in research consortia formed by the government.

It is therefore clear that in Canada it matters in which province the research is carried out. This is documented in Table 31 which shows that the cost of the research dollar can be as low as 40 cents, not counting direct subsidies, of course. An international comparison of tax and subsidy régimes will be undertaken in the next chapter.

Click here to view Table 31: After-tax Costs of R&D Expenditures in Ontario and Quebec, 1990

Normatively Correct Subsidies to Innovativeness

At the start we should remind ourselves that while most of the literature speaks of R&D subsidization, we know that the economically interesting issue is the support to innovativeness: both the creation and the diffusion of new products and processes. The alleged market failure that may necessitate the remedy of subsidy occurs with respect to innovativeness, not with respect to R&D.

We have already discussed at length the rationales for public intervention under the three headings of market failure and in respect of strategic trade considerations in chapter 2. Strategic trade policy with respect to innovation is hard to discern in Canada and nobody really knows what to prescribe for it, despite recent advances in game theory. Similarly, nobody knows what is a critical level of risk, or a minimum efficient size for innovative behaviour so that a policy response can be triggered.

We are thus left in what is, in any case, the most discussed form of alleged failure, namely inappropriability, as a reason for public policy intervention. In other words, it may be that the inability to appropriate "sufficient" returns from innovation activity inhibits the firm from carrying out a "socially optimal" level of such an activity.

Inappropriability is not, in common with many economic concepts, easy to define in operational terms. Expenditures on basic research are most often an investment in inappropriable But not always. See the discussion and reference to Mowery and Rosenberg in Chapter 3.knowledge. A percentage of patents originating in industry i and most likely to be used in other industries j is some indication of spillover. An ambitious U.S. survey of industry, asking questions such as "How effective are patents in your industry to prevent competitors to duplicate a new product?" and "What is the average time to duplicate a patented, major product innovation?" provides numerical proxy variables for the level of inappropriability in various Richard C. Levin et al., "Appropriating the Returns from Industrial R&D," Brookings Papers on Economic Activity, 1987, 3, pp. 783-831.industries.

But inappropriability can often be gauged in individual, specific cases of innovative products or processes-it is the equivalent of the consumer surplus that is generated by it. A graphical depiction of increase in that surplus was shown in Figure 8, panel (b) as the diagonally striped area P1P2R1R2. (Consumer surplus was defined as the difference between the unit price each customer would have been willing to pay for the product and the actual price, multiplied by the relevant quantities purchased). Reasonably simple ways of estimating that surplus include asking questions as to how much an existing product's price could be lowered due to a new lower-cost process-and so how many more units could be The Canadian authority on this subject is Abraham Tarasofsky's The Subsidization of Innovation Projects by the Government of Canada, Ottawa: The Economic Council of Canada, 1984.sold.

The prescription, then, is to offer the potential innovator a subsidy which would cover the gap between the present value of his future innovation outlays and the present value of future receipts from the innovation, using his customary rate of discount. Provided, of course, that the subsidy shall not exceed the present value of the consumer surplus generated, less costs of granting the subsidy. A handy formula summarizes the prescription for S, the subsidy:

g - G = S £ B - G - C

where g is the private cost of the innovation outlay, G its private benefit, B the economic benefits to society (consumer surplus) from the project and C the cost of the subsidy's delivery; all variables deemed in present Tarasofsky, op. cit., p. 17. values.

An equivalent way of approaching the determination of optimal subsidy has been outlined by Dan Usher, The Benefits and Costs of "Firm-Specific" Grants: A Study of Five Federal Programs, Queen's University, Dept. of Economics Paper, Jan. 1, 1983.Usher. The projects will be worthy and the subsidies necessary when the claimants and grantors can show that three incremental conditions are satisfied:

1.The project is incremental to the firm, that is, the applicant firm must document, and the grant administrator verify, that the firm's project will cost more than the present value of its expected private benefit.

In more mundane terms, the firm must make risk adjusted projections, properly discounted, of the project's future flows of costs and revenues, much as it would undertake for any other capital investment. A negative private return, a deficit, would imply that a subsidy is needed. At the same time this calculation would indicate the amount of the subsidy, which otherwise-while incremental-may be excessive.

2.The project is incremental to the market in addition to being incremental to the firm, that is, there is no other firm that could have undertaken the project profitably without a subsidy.

The fulfilment of this condition requires some investigation of the innovative, or at least R&D, activities of current or potential competitors.

3.The project is also incremental to the economy, that is, the benefits accruing to the economy as a consequence of the innovation must be sufficient to offset the subsidy granted and the cost of its delivery.

Practically this means that the social benefits of the grant must be calculated by the government agency and diminished by the direct cost of the subsidy and its direct costs of both application and administration. If the remainder is positive, the grant is incremental to the economy.

While the private and public benefits can, in the opinion of experts, be calculated reasonably easily, there is greater uncertainty surrounding C, the cost of transferring the funds from taxpayers to subsidies. What is known is that such costs are much higher than the general public is aware of. Tarasofsky estimates them as Op. cit., p. 13.follows:

Click here to view Table

The consensus among public finance economists as to the surprisingly high marginal deadweight costs occasioned by small tax increases is as yet little known. The concept of the deadweight tax loss cannot be explored here, but good references The basic reference is Arnold C. Harberger, "Taxation, Resource Allocation and Welfare," in J. Due (ed.), The Role of Direct and Indirect Taxes in the Federal Revenue System, Princeton: Princeton University Press, 1964.exist.

The conditions enumerated previously seem logical, eminently reasonable, and what is more, practically attainable. Have they guided the federal innovation grant programs? Tarasofsky, who examined in detail the three programs listed as largest during the late '70s and early '80s (in table 29), replied "no" quite Tarasofsky, op. cit.emphatically.

With regard to DIPP Tarasofsky found it impossible to measure, on the basis of the information available to the program's administrators, the inappropriable benefits accruing to foreign or domestic Since DIPP is heavily export oriented, Canadian taxpayer-financed innovation would presumably also increase foreign customer consumer surplus.markets. In respect of IRAP (but not PILP, merged with IRAP since 1986, a program aimed at technology diffusion from federal labs to industry), Tarasofsky stated that "the information formally required from the applicants is utterly incapable of permitting a rational judgement as to whether the project warrants Tarasofsky, op. cit., p. 59.subsidization." Finally, regarding the EDP grant program, Tarasofsky said that its most striking administrative shortcoming lay in the failure to recognize the relevance (and to provide for the projection) of the proposed projects' inappropriable Op. cit., p. 40.benefits. The assiduous reader may wish to check the most recent criteria used by ISTC and the National Research Council for grants under DIPP and IRAP-R programs in the CCH Canadian Op. cit., IRAP-R p. 68, DIPP, p. 175.handbook. Hanel and Palda have looked at the grant programs in a more aggregated and summary Petr Hanel and Kristian Palda, "Appropriability and Public Support of R&D in Canada," Prometheus, December 1992, pp. 204-226.way. Using proxies for appropriability they found that the IRDP grant-receiving industries did tend to have difficulties in appropriating returns for R&D through patent protection.

Thus far the appropriability question has focused on grants. Since tax incentives, almost by definition, apply to broad classes of corporate taxpayers, inappropriability plays hardly any role in them. And because, as was seen in Table 28, tax credits have become by far the largest part of R&D subsidization, the theoretically correct approach to the stimulation of research in Canada is increasingly in abeyance. But even this theoretically correct approach is now being questioned, as will be pointed out in the concluding chapter, and further conditions for subsidization are invoked.

We made the point that most discussion revolves around subsidies to R&D rather than to innovativeness. Clearly the rules with respect to innovation are not different from those with regard to R&D: one merely puts the accent on all expenditures, not just R&D, that go into the launching of a new process, such as new plant and distribution facilities, prior market research, marketing launch outlays, and so on. There are some subsidy programs in Canada that take partial account of such "R&D complementing" expenses. Among them are DIPP, the Atlantic Canada Opportunities Agency Action Program (ACOA), and the Federal-Quebec EDP program.

The tax system, as already mentioned, offers preferential treatment strictly to R&D expenditures. As far as this author could ascertain from discussions with ISTC officials, Canadian corporate taxation managers, American experts, German and Japanese embassy specialists, and from chartered accounting firms' internationally-oriented reports, tax régimes in these other three countries are no different-only R&D expenses "qualify." (On this point, for example, see the exchange of faxes between the writer and the Japanese Institute of Science and Technology in Appendix 6A). A difference between Canadian and foreign R&D tax regimes is pointed out in Box 8.

Click here to view Box 8: A Shortfall in Canada's R&D Tax Generosity

A Word on Risk, MES and Diffusion

As yet, there have been no hard and fast normative prescriptions devised to advise subsidizers about risk, MES (minimum efficient scale), and diffusion. But some useful observations can be made.

The words "risk" or "riskiness" appear as partial criteria for grant approval in several programs (IRAP-R, DIPP, Quebec Industrial Development Corporation's research and innovation activities). Circular 86-4R2 (August 29, 1988) of Revenue Canada Taxation states, however, that "the scientific or technological uncertainty, rather than the economic and financial risk is important in characterizing scientific research and experimental development-and hence eligible activities" (p.6). And of course, as we have seen, the tax law is not inclined to extend favourable treatment beyond strictly R&D to include other innovation-connected expenses.

Figure 13 shows a well known time pattern of profit flows accompanying the research, development and launching of a new product. As drawn, it is expected that the drain on corporate profits will be largest not during the R&D phase, but in the initial manufacturing and marketing launch stage. It is perhaps at point R that the risk of the venture may appear to be the heaviest to investors. Yet there is no tax provision to alleviate the burden of risk at this moment in the new product's life, no tax credit for at least a part of the new manufacturing costs or for part of the marketing investment that trains sales representatives and diffuses information to customers.

Click here to view Figure 13: Typical Sales and Profits of a New Product

Figure 14 is a slightly different, but real-life, illustration of the investment burden accompanying a significant new-product launching. The U.S. National Research Council proposes that net cash flows take more than 10 years before turning positive in a typical medium-size-and-range aircraft project, carried from the research stage to the delivery of 700 planes. Canada's income tax provisions allow tax credits strictly for R&D expenditures. What the preceding analysis suggests is that because the maximum risk of an innovative undertaking accompanies outlays that take place beyond the research stage, some provisions for innovation after R&D outlays should be made in the tax credit scheme.

Click here to view Figure 14: Typical Cash Flow Curve for Large Transport Aircraft Program

As regards the MES, or minimum efficient size for a research undertaking (lab, engineering facility, software design team, etc.), there are two provisions currently and generously in place. Small firms which will never have a hope of undertaking research are amply served by federal and provincial ministry research facilities. Agriculture Canada spent over $325 million on R&D in 1990-91, Fisheries and Oceans around $135 million, Forestry Canada over $50 million. The National Research Council's primary mission is to help industry with research. For instance it spends in-house tens of millions of dollars on building research.

The second way by which alleged MES shortfalls are overcome is the federal tax credit (35 percent for CCCPs, only 20 percent for others, as already mentioned) and the various provincial tax provisions. Finally, many subsidy-granting programs have an in-built bias or open commitment to small and medium businesses.

Finally, diffusion of new technologies is by now probably in the forefront of both federal and provincial policies in support of innovation, as was already stated in chapter 2 in respect of the grants programs listed in table 3. First and foremost is the National Research Council, with the largest federal research budget-ahead of Agriculture Canada whose integrated task is both to invent and to diffuse-of which a considerable part is devoted to technology transfer:

NRC's Industrial Research Assistance Program (IRAP) funds firms to adopt new technologies. In 1988, IRAP was a source of technical assistance for roughly 5,000 firms. IRAP seeks the technology needed by a firm and transfers it by any appropriate means. The needed technology may be obtained from government laboratories, universities, technological institutes, other firms or foreign sources...

NRC, through the Canadian Institute for Scientific and Technological Information (CISTI), offers a renowned S&T information collection, dissemination and delivery service. CISTI responded to some 425,000 requests for material last year, of which about half were in support of ISTC, Strategic Overview, op. cit., pp. 18-19.industry...

Other federal ministries, such as Energy Mines and Resources Canada, with its CANMET laboratories, are heavily engaged in Margot J. Wojciechowski, Research and Development in the Mineral Sector, Ottawa: EMR-CANMET Report CM 89-2E, August 1989.transfer. Industry Science and Technology Canada funds grant programs to support diffusion, such as the Strategic Technologies Program ($17 million in 1990-91), Technology Outreach Program ($18 million), Advanced Manufacturing Technology Program ($2 million), all listed in table 3, and the Technology Inflow Program for the acquisition of foreign technology. More detailed descriptions of transfer/diffusion programs are available in the already cited ISTC's Strategic Overview of S&T Activities in the Federal Government 1989-90 and in National Research Council's annual reports. The preoccupation of the federal government with diffusion comes through via reports of the National Advisory Board on Science and Technology (NABST), such as the one of November 2, Revitalizing Science and Technology in the Government of Canada, Section on technology transfer, pp. 38-45.1990. That report quotes with admiration and envy the U.S. Federal Technology Transfer Act of 1986 which is designed to facilitate diffusion from government labs to the private sector.

Among the provinces, Ontario, for instance, is very active in the diffusion of technology. The Ontario Technology Fund in its annual 1989-1990 report states that the province spent between 1987/88 and 1989/90 $97 million on the establishment of so-called Centres of Excellence and committed $204 million to See also footnote 43 in Chapter 2.them. Among their three objectives is the encouragement of the transfer and diffusion of technology to industry. The centres are located at universities, such as the Ontario Laser and Lightwave Research Centre at the University of Toronto, and the Ontario Centre for Materials Research, based at Queen's University.

We are thus led to the conclusion that the diffusion side of innovativeness receives very likely as generous a support from the Canadian taxpayer as the innovation side. A doubt, previously emitted with respect to innovation, applies as well to its diffusion. It is not clear why efficiently-managed firms in the private sector would not find it in their interest to inform themselves about ongoing technological advances or why they would not band together and insist that their associations provide them with this information. If they do not do this, then they either are not efficiently managed or they respond rationally to conditions in the environment which are not conducive to early adoption. In the first instance, should government really believe that it can improve managerial skills, at least in the private sector? In the second instance, should government intervene to alter the environment such as to increase competition through tariff reduction or to manipulate, for example, interest rates in order to ensure greater stability-and so long-term technological investment-in the construction sector? This type of argument does not, naturally, apply to areas of the economy (health, public administration, etc.) where innovation incentives do not operate.

How Effective has Government Support Been?

This study has looked at the formidable array of grant programs and tax stimulants to R&D, innovation, and diffusion at the federal and provincial levels. The next step is to ask of what effect they have been. Clearly, this is too ambitious a question and one that can only be answered piecemeal and incompletely. The answers can be broken down into two categories-the effects of tax stimulants and the effects of grants. They are not uniform, even within each category.

Several authors have examined the impact of Canadian tax policies. In his 1985 contribution to the Macdonald royal commission, Bernstein came to the conclusion that tax incentives generally lead to a dollar-for-dollar increase in R&D expenditures. A year later Bernstein scaled his estimation down to an 80 cent increase per dollar of tax Jeffrey I. Bernstein, "Research and Development, Patents, and Grant and Tax Policies in Canada," in D.G. McFetridge (ed.), Technological Change in Canadian Industry, Toronto: University of Toronto Press, 1985, pp. 1-41. See also Bernstein, "The Effect of Direct and Indirect Tax Incentives on Canadian Industrial R&D Expenditures," Canadian Public Policy, September 1986, pp. 438-48.incentive. In that same year Mansfield and Switzer estimated, on the basis of a survey sample of 110 American, 55 Canadian and 40 Swedish companies, that both tax allowances and tax credits increased R&D expenditure by about 30 to 40 cents for each dollar of tax E. Mansfield, and L. Switzer, "The Effects of R&D Tax Credits and Allowances in Canada," Research Policy, 1985, 14, pp. 97-107.abatement. Writing a year later, Mansfield pointed out with reference to his previous study that:

In all cases, the increased R&D expenditures due to tax incentives seemed to be substantially (and significantly in a statistical sense) less than the revenue lost by the government...moreover, in each country, there was substantial evidence that these tax incentives resulted in considerable redefinition of activities as R&D...of about 13 to 14 percent in both Canada and Edwin Mansfield, "The R&D Tax Credit and Other Technology Policy Issues," American Economic Review, May 1986, 190-94. See also Mansfield and Switzer, "How Effective are Canada's Direct Tax Incentives?," Canadian Public Policy, 1985, 11, pp. 241-46.Sweden.

What of the grant form of subsidies? We can ask, for instance, whether the R&D expenditures of private companies tend to increase (over what they would have been otherwise) as a result of receiving government subsidies.

In one of the best-known articles about R&D in Canada, Howe and McFetridge raised the question as to whether a direct government grant to a manufacturer acts as seed money for other R&D projects, or whether it proves merely to be a transfer J.D. Howe and D.G. McFetridge, "The Determinants of R&D Expenditures," Canadian Journal of Economics, February 1976, pp. 57-71.payment. The authors had access to data assembled by the Department of Industry, Trade and Commerce which funded several incentive grants programs.

Incomplete from Page 169

info@fraserinstitute.ca You can contact us at the above email address for any comments or information requests. Please report any dead links or technical problems.

If you know someone who would be interested in this web page, please enter their email address below, and we will forward this URL to them: Email Address:

Last Modified: Wednesday, October 20, 1999.