Innovation and Growth from a Schumpeterian Perspective

INNOVATION AND GROWTH FROM A SCHUMPETERIAN PERSPECTIVE

Dalloz | « Revue d'économie politique »

2018/5 Vol. 128 | pages 693 à 711 ISSN 0373-2630DOI 10.3917/redp.285.0693

Article disponible en ligne à l'adresse :--------------------------------------------------------------------------------------------------------------------https://www.cairn.info/revue-d-economie-politique-2018-5-page-693.htm--------------------------------------------------------------------------------------------------------------------

Distribution électronique Cairn.info pour Dalloz.© Dalloz. Tous droits réservés pour tous pays. La reproduction ou représentation de cet article, notamment par photocopie, n'est autorisée que dans leslimites des conditions générales d'utilisation du site ou, le cas échéant, des conditions générales de lalicence souscrite par votre établissement. Toute autre reproduction ou représentation, en tout ou partie,sous quelque forme et de quelque manière que ce soit, est interdite sauf accord préalable et écrit del'éditeur, en dehors des cas prévus par la législation en vigueur en France. Il est précisé que son stockagedans une base de données est également interdit.

Powered by TCPDF (www.tcpdf.org)

© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T

éléchargé le 26/08/2022 sur ww

w.cairn.info (IP

: 65.21.229.84)

https://www.cairn.info/revue-d-economie-politique-2018-5-page-693.htm

http://www.tcpdf.org

Innovation and Growth from a SchumpeterianPerspective

Philippe Aghion1

In this lecture we argue that important aspects of the growth process cannot easily beaccounted for using models where capital accumulation is the main source of growth.The four aspects we emphasize in this lecture are: the transition trap, secular stagna-tion, the recent rise in top income inequality, and firm dynamics. The lecture arguesthat by contrast these aspects can be addressed by the Schumpeterian growth para-digm in which: (i) growth results primarily from innovation; (ii) innovation responds toincentives shaped by economic policies and institutions; (iii) new innovations replaceold technologies (creative destruction).

creative destruction – R&D – entry – exit – competition – technology frontier – firm

dynamics

Innovation et croissance : une perspectiveschumpetérienne

Cette leçon présidentielle vise à montrer que des aspects importants du processus decroissance économique sont difficiles a appréhender a travers des modèles où lasource principale de la croissance est l’accumulation du capital physique. Quatreaspects sont discutés, en particulier : les trappes de non-transition, la stagnation sécu-laire, la relation entre croissance et inégalités de revenus, et la relation entre croissanceet dynamique des firmes. Cette leçon montre que ces phénomènes peuvent aucontraire être analyses a l’aide du paradigme schumpétérien dans lequel : (i) la crois-sance résulte au premier chef de l’innovation ; (ii) l’innovation répond à des incitationsconstruites sur les politiques économiques et les institutions ; et enfin (iii) les innova-tions d’aujourd’hui remplacent les innovations d’hier via un processus de destruction-créatrice.

destruction créatrice – R&D – innovation – concurrence – frontière technologique –

dynamique des firmes

JEL Classifications : O10, O11, O12, O30, O31, O33, O40, O43, O47

1. Collège de France, 3 Rue D’Ulm 75005 Paris France and London School of Economics,UK. Email: [email protected]

•L

XV

IeC

ON

GR

ÈS

AN

NU

EL

DE

L’A

FSE

,20

17

REP 128 (5) septembre-octobre 2018

© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

1. Introduction

Malinvaud’s celebrated 1953 article on “Capital Accumulation and the Effi-cient Allocation of Resources” introduced the notions of capital, interestrates and dynamic efficiency into the Arrow-Debreu general equilibriumframework. This was a major step which allowed to both generalize and alsoclarify previous work on capital theory (e.g. by Allais, Bohm-Bawerk, andWicksell), and at the same time a milestone for forthcoming neoclassicalgrowth models (by Solow, Ramsey, Cass-Koopmans) based on capital accu-mulation. That these models in turn have served as a benchmark for growtheconomists over the past three decades, is largely due to their having inher-ited from (or shared with) Malinvaud three concerns: namely, parcimony inmodeling and assumptions, and clarity and rigor in the analysis.

Yet, while taking Malinvaud as a model and benchmark, in this paper weshall argue that important aspects of the growth process cannot easily beaccounted for using models where capital accumulation is the main sourceof growth. The four aspects we consider are: the middle income trap, secularstagnation, the recent rise in top income inequality, and firm dynamics. Thepaper argues that by contrast these aspects can be addressed by the Schum-peterian growth paradigm in which: (i) growth results primarily from inno-vation; (ii) innovation responds to incentives shaped by economic policiesand institutions; (iii) new innovations replace old technologies (creativedestruction).

The remaining part of the paper is organized as follows. First, we will askthe question: what do we need more than a theory of economic growthbased on capital accumulation. Second, we shall present the Schumpeterianparadigm. Third, we shall discuss the four aspects on which the Schumpe-terian approach sheds new light compared with growth models based oncapital accumulation: the middle income trap, secular stagnation, the recentrise in top income inequality, and firm dynamics.

2. The limits of growth theories basedon capital accumulation

During my student years, the dominant paradigm in growth economicswas the neoclassical growth models, which were taught first under theassumption of a constant savings rate (the Solow model) and then in moreadvanced courses in the context of an economy where a representativeconsumer decides about consumption, savings and investment by maximiz-ing her intertemporal utility (the Ramsey-Cass-Koopmans model). Eventhough these models were directly linked to Malinvaud’s seminal 1953paper, his paper was mentioned as a reference but not often taught in class,I guess mainly because of its higher degree of complexity.

694 ———————— Innovation and Growth from a Schumpeterian Perspective


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

The Solow model was indeed particularly simple to teach, in which thewhole dynamics of the economy was described in two equations. But some-what paradoxically, a main conclusion of this growth model with capitalaccumulation was to show why there can be no long run growth withoutsomething more than capital accumulation, namely technical progress.

I will not bother you here with the details of this model which economistsall know too well. But in a nutshell: the model describes an economy wherefinal output is produced using capital as input, and where therefore it is theaccumulation of capital which generates output growth. This corresponds tothe first equation of the model. Then the question is: where does capitalaccumulation come from? This in turn is answered with the second equationof the model: from savings (aggregate savings equal aggregate investmentin equilibrium) and savings in the Solow model are a constant fraction offinal output (i.e. of aggregate GDP).

One might think that everything should go well in such an economy: morecapital stock financed by savings will produce more final output, which willtranslate into more savings (as savings are proportional to final output) andtherefore in still more capital stock and so on.

The problem is that we run in to decreasing returns when trying toincrease output by increasing the capital stock: the higher the existing stockof capital (number of machines), the lower the marginal increase in outputfrom increasing the stock of capital by one unit (i.e. from adding one moremachine). Thus the lower the increase in savings and therefore the lower theinduced increase in capital stock.

At some moment, the process of capital accumulation runs out of steam(it stops when capital depreciation catches up with marginal savings) atwhich point the economy stops growing. To generate sustained long-termeconomic growth, there must be continuous technical progress to increasethe quality (productivity) of machines. But Solow does not tell us wheretechnical progress is coming from.

In addition, if the model predicts conditional convergence, it does not giveus the tools to understand why some countries converge to the standards ofliving (per capita GDP) of developed countries whereas other countries do notconverge; or why some countries with lower capital stocks grow less rapidlythan other countries with higher capital stocks, or why capital does not neces-sarily flow from rich to poor countries (the so-called Lucas Paradox).

More fundamentally, the model does not look at growth from the point ofview of firms and entrepreneurs: which economic environment or “businessclimate” (institutions, policies) is more likely to stimulate innovation andentrepreneurship?

These motivated us to elaborate a new paradigm, where growth resultsprimarily from quality-improving innovations.

Philippe Aghion —————————————————————————————————————————————————————— 695


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

3. The Schumpeterian paradigm

In the fall of 1987 Peter Howitt and I elaborated a growth model revolvingaround three important ideas outlaid by the Austrian economist JosephSchumpeter (Aghion and Howitt [1992]).

First idea: long-run growth is primarily generated by innovations (this isthe natural counterpart of Solow’s conclusion that no long-run growth canbe expected without sustained technological progress).

Second idea: innovations result from entrepreneurial investments (R&D,training, computer purchase, etc.) and entrepreneurs respond to the eco-nomic incentives (positive or negative) which result from economic policiesand economic institutions. Thus typically innovation-based growth will bediscouraged in environments with poor property right protection or withhyperinflation as these will damage the profitability from innovation. Inother words, innovation-based growth is a social process and we can talkabout policies of growth and institutions of growth.

Third idea: creative destruction. New innovations replace old technolo-gies, Schumpeterian growth is a conflictual process between the old and thenew: it tells the story of all these incumbents firms and interests whichpermanently try to prevent or delay the entry of new competitors in theirsector. Hence there is something called “the political economy of growth”.

A distinct prediction of the Schumpeterian growth model is that firm orjob turnover should be positively correlated with productivity growth.Another distinctive implication of the model is that innovation-led growthmay be excessive under laissez-faire. Growth is excessive (resp. insufficient)under laissez-faire when the business-stealing effect associated with creativedestruction dominates (resp. is dominated by) the intertemporal knowledgespillovers from current to future innovators.

After an initial period of excitement with this new paradigm, some empiri-cal studies came out which tempered our enthusiasm.

In particular Blundell et al. [1995, 1999], using UK firm-level data, raiseddoubt on another prediction of our model: namely the prediction that morecompetition should be detrimental to growth by reducing monopoly rentsfrom innovation and thus entrepreneurs’ incentives to invest in innovation inthe first place as shown in Figure 1 (incidentally, this latter argument hasbeen used by Bill Gates when facing anti-trust action). Instead, when per-forming a linear regression of firms’ innovation intensity and/or productivitygrowth on the degree of product market competition in the firms’ sectors,Blundell et al. found a positive correlation between competition andinnovation/growth, as shown in Figure 2.

How, if at all, could we reconcile theory and evidence? Should we throwthe model in the garbage bin and start again from scratch? Or should wesimply ignore the empirical challenges and proceed as before?

We went for a third way: namely to look more closely at the model and tryto identify the assumption or assumptions which generate this counterfac-tual prediction of a negative relationship between competition and growth.



© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

And we finally identified the culprit: in our initial model only currentlyinactive firms innovate, not the currently active firms (i.e. not the currenttechnological leaders). Thus an innovating firm in our model would movefrom zero profit (pre-innovation) to a positive profit (post-innovation). Thenno wonder why competition would discourage innovation: competitionreduces the post-innovation profit which here is equal to the net profit frominnovation.

Fig. 1 Competition and growth: theoretical prediction

Fig. 2 Competition and growth: empirical relationship

Philippe Aghion —————————————————————————————————————————————————————— 697


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

However in reality one finds at least two types of firms in most sectors ofthe economy and these two types of firms do not react in the same way toincreased competition. You first have what we call “frontier firms”, i.e. firmsthat are close to the current technological frontier in their sector. These firmsare currently active and they make substantial profits even before innovatingthis period. Second, you have what we call the “laggard firms”, i.e. firms farbelow the current technological frontier. These firms make low profits andtry to catch up with the current technology frontier.

To try and understand why these two types of firms react differently tocompetition, imagine for a moment that what you are looking at are notfirms but students in a classroom. And among them you have the top stu-dents and the bottom of the class. And suppose that you are opening theclass to an additional student who turns out to be a very good student. Thisis how I represent an increase in competition in this context. How will thestudents react to this new student joining the classroom? The answer (hereI refer to important work by Caroline Hoxby who studied precisely this) isthat letting the new student in will encourage the other top students to workharder in order to remain the best, whereas it will further discourage stu-dents at the bottom of the class, as those will find it even harder to catch up.

Quite strikingly, firms react like classroom students: namely, faced with ahigher degree of competition in their sector, firms that are close to thetechnology frontier will innovate more in order to escape competition,whereas firms that are far from the technological frontier and try to catch upwill be discouraged by the higher degree of competition, and as a resultinnovate less: these latter firms behave like in the basic Schumpeterianmodel (see Figure 3).

Fig. 3 Competition, growth and distance to frontier



© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

Overall, the effect of competition on innovation and productivity growth isan inverted-U, as shown in Figure 4, which synthetizes the positive escapecompetition effect and the negative discouragement effect. The prediction ofopposite reactions of frontier versus non-frontier firms to competition, andof an inverted-U overall, were tested and confirmed in joint work with Rich-ard Blundell, Nick Bloom and Rachel Griffith using the same kind of firm-level data as in the empirical studies I mentioned above.

In the end, this exercise has been mutually enriching. On the one hand,our empirical colleagues realized that the relationship between competitionand growth was more involved and subtle than what they thought based ontheir initial studies. On the other hand, we understood how to enrich ourmodel so as to bring out, not one but two basic effects of competition oninnovation and growth, to identify conditions under which one or the othereffect dominates, and why when aggregating across all firms/sectors weobtain an inverted-U relationship, as anticipated by Scherer [1967].

To reconcile theory with evidence we extended our basic Schumpeterianmodel by allowing for step-by-step innovation in the Schumpeterian growthmodel (see Aghion et al. [1997]; Aghion et al. [2001]). Namely, a firm that iscurrently behind the technological leader in the same sector or industrymust catch up with the leader before becoming a leader itself. This step-by-step assumption implies that firms in some sectors will be neck-and-neck. Inturn in such sectors, increased product market competition, by making lifemore difficult for neck-and-neck firms, will encourage them to innovate inorder to acquire a lead over their rival in the sector. This we refer to as theescape competition effect. On the other hand, in unleveled sectors wherefirms are not neck-and-neck, increased product market competition will tendto discourage innovation by laggard firms as it decreases the short-run extraprofit from catching up with the leader. This we call the Schumpeterianeffect. Finally, the steady-state fraction of neck-and-neck sectors will itselfdepend upon the innovation intensities in neck-and-neck versus unleveledsectors. This we refer to as the composition effect.

Fig. 4 Competition and growth: the inverted-U relationship

Philippe Aghion —————————————————————————————————————————————————————— 699


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

This extended model predicts that in the aggregate the relationshipbetween competition and innovation should follow an inverted-U pattern.Intuitively, when competition is low, innovation intensity is low in neck-and-neck sectors, therefore most sectors in the economy are neck-and-neck (thecomposition effect); but precisely it is in those sectors that the escape com-petition effect dominates. Thus overall aggregate innovation increases withcompetition at low levels of competition. When competition is high, innova-tion intensity is high in neck-and-neck sectors, therefore most sectors in theeconomy are unleveled sectors, so that the Schumpeterian effect dominatesoverall. This inverted-U prediction is confirmed by Aghion et al. [2005],using panel data on UK firms.

The prediction that more intense competition enhances innovation in“frontier” firms but may discourage it in “non-frontier” firms, was tested byAghion et al. [2009a] using again panel data of UK firms.

Another prediction from our extended model, is that there is complemen-tarity between patent protection and product market competition in foster-ing innovation. Intuitively, competition reduces the profit flow of non-innovating neck-and-neck firms, whereas patent protection is likely toenhance the profit flow of an innovating neck-and-neck firm. Both contributeto raising the net profit gain of an innovating neck-and-neck firm; in otherwords, both types of policies tend to enhance the escape competition effect.

That competition and patent protection should be complementary inenhancing growth rather than mutually exclusive is at odds with both, ourfirst model and Romer [1990], where competition is always detrimental toinnovation and growth (as we discussed above) for exactly the same reasonthat intellectual property rights (IPRs) in the form of patent protection aregood for innovation: namely, competition reduces post-innovation rents,whereas patent protection increases these rents. But it is also at odds withBoldrin and Levine [2008] who hold that patent protection is always detri-mental to innovation and growth in their model where competition is goodfor growth.

Our prediction of a complementarity between competition and patent pro-tection, was tested by Aghion et al. [2013] using OECD country-industrypanel data.

4. Four aspects of economic growth

The Schumpeterian growth theory and the firm-based empirical analysis itgenerated, can shed light on important aspects of the growth process. Herewe consider four aspects: (i) the middle-income trap; (ii) secular stagnation;(iii) the dynamics of inequality; (iv) firm dynamics.



© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

4.1. The transition trap

How can explain that some countries grow fast for a while and therebycatch (at least to some extent) with more advanced countries, but then atsome point they stop converging and enter periods of low growth? Oneexample in point is Argentina. In 1890, Argentina enjoyed a GDP per capitaapproximately 40% that of the United States, which made it a middle-income country. This level was three times the GDP per capita of Brazil andColombia and equivalent to that of Japan at the time. Argentina sustainedthis level of 40% of the GDP per capita of the United States through the1930’s. To be precise, Chow’s test (a statistical test) shows a break around1938, after which Argentina’s productivity declines relative to American pro-ductivity by approximately 21% per year. What explains this drop-off?

More recently, Japan has experienced very fast growth between WWII andthe late 1980, but ever since it has experienced very sluggish growth. Some-what similarly, France experienced fast growth between 1945 and 1973 (theso-called *Trente Glorieuses*) but then growth has slowed down.

Schumpeterian growth theory offers the following explanation. Somecountries either had institutions or had implemented policies (in particularimport-substitution) that fostered growth by accumulation of capital andeconomic catch-up. They did not, however, adapt their institutions to enablethem to become innovating economies. As demonstrated in joint work withDaron Acemoglu and Fabrizio Zilibotti, the greater the level of developmentin a country —i.e., the closer it gets to the technology frontier— the greaterthe role of cutting edge innovation as the motor of growth, replacing accu-mulation and technological catch-up (Acemoglu et al. [2006]).

Japan, where the State has always tightly controlled competition, is anillustrative example: Japan’s Ministry of Economy, Trade and Industry (MITI)caps the number of import permits, and the State subsidizes investment bythe big industrial-financial consortia known a keiretsu. It is thus not surpris-ing that from an extremely high level between 1945 and 1985 —the envy ofother developed countries— Japan’s growth has fallen to a very low levelsince 1985.

In the previous section we mentioned some recent evidence for the pre-diction that competition and free-entry should be more growth-enhancing inmore frontier firms, which implies that they should be more growth-enhancing in more advanced countries since those have a larger proportionof frontier firms. Similarly, Acemoglu et al. [2006] show, using a cross-country panel of more than 100 countries over the 1960–2000 period, that:

1. Average growth should decrease more rapidly as a country approachesthe world frontier when openness is low.

2. High entry barriers become increasingly detrimental to growth as thecountry approaches the frontier.

Philippe Aghion —————————————————————————————————————————————————————— 701


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

These two empirical exercises point to the importance of interacting insti-tutions or policies with technological variables in growth regressions: open-ness is particularly growth-enhancing in countries that are closer to thetechnological frontier; entry is more growth-enhancing in countries or sec-tors that are closer to the technological frontier.

Similarly, to the extent that frontier innovation makes greater use ofresearch education than imitation, the prediction is that the more frontier aneconomy is, the more growth in this economy relies on research education.Aghion et al. [2009b] showed that research-type education is always moregrowth-enhancing in US states that are more frontier, whereas a biggeremphasis on two-year colleges is more growth-enhancing in US states thatare farther below the productivity frontier.

4.2. The debate on “Secular Stagnation”

In 1938, economist Alvin Hansen explained in his Presidential Addressbefore the American Economic Association that in his opinion, the UnitedStates faced inexorable weak growth in the long term (Hansen [1939]). Thenation was just emerging from the Great Depression, and Hansen did notanticipate another World War that would stimulate a rebound in publicspending and thereby of aggregate demand.

Since then, we have experienced another major financial crisis, the 2007crisis, which led Larry Summers and others to revive the expression “secu-lar stagnation” to characterize a situation that they assimilated to the onedescribed by Hansen in 1938 (Summers [2013]). Summers’s argument is thatinvestment demand was so weak that negative interest rates were necessaryfor a return to full employment.

Robert Gordon however, believes that the risk of secular stagnationreflects a supply problem (Gordon [2012]). Gordon proposes that the age ofgreat innovations is past. He uses the metaphor of a fruit tree: the low-hanging fruit is the best; after that the fruit is harder to pick and less juicy.

By way of example, the arrival of the Boeing 707 on the market in 1958marked the end of progress in duration of air travel time. Until then, traveltime had decreased exponentially; since then, not only has travel timeceased to decrease, it has actually increased due to the conjugated effects ofenergy-saving and cost-optimization.

Schumpeterian economists are more optimistic about the future thanSummers and Gordon. A first argument (Jorgenson) is that the revolution ininformation and communications technologies (ICT) has radically and dura-bly improved IT-producing technology; meanwhile globalization (which wasconcomitant with the ICT revolution) substantially increased the potentialreturns on innovation —the scale effect— as well as the potential downsideof not innovating— the competition effect.



© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

A second argument against the secular stagnation view, is that we havewitnessed an acceleration in innovation over the last several decades, whichhas been fully reflected by measured productivity growth.

In particular, Aghion et al. [2018] argue that innovation involving creativedestruction is not properly taken into account by current measures of TFPgrowth. Whenever old products in the PPI are replaced by new products bynew entrants, the statistical office often uses the price changes of survivingproducts to infer the price change of the replaced products. This is refereedto as imputation.

Using the Schumpeterian growth paradigm to provide explicit expressionsfor missing growth from creative destruction, they estimate a missinggrowth from imputation in the US equal to about 0.6 percentage point peryear on average over the past thirty years. Moreover, they show that missinggrowth has not declined over time. Finally they show that most of the miss-ing growth from imputation is associated with creative destruction, i.e. fromnew products replacing old products, rather than from variety expansion.

Follow-up work by Aghion et al. [2018] shows a missing growth fromimputation equal to 0.5 percentage points over the 2006-2013 period;although this is slightly smaller than the missing growth from imputation inthe US, missing growth from imputation in France represents a higher frac-tion of total TFP growth. Moreover, Aghion et al. [2018] show that missinggrowth is higher in regions or sectors with higher degrees of creativedestruction. Finally, they show that the similarity between missing growth inthe US and in France hides significant differences in the employment sharesof new entrants and of exiting firms between the two countries: those sharesare higher in France than in the US; however, since missing growth can beshown to proportional to the difference between the employment share ofnew entrants today and the employment share of exiters yesterday, the factthat these two employment shares is bigger in France than in the US by thesame amount, implies that the difference between these employmentshares, and therefore missing growth from imputation, is similar betweenthe two countries.

The following figure depicts the relationship between the level of creativedestruction in a sector at a particular time on the one hand, and the corre-lation between TFP growth and the annual flow of patents on the otherhand. More precisely: (i) each year over the period from 1994 to 2010, werank 16 sectors, spanning the whole manufacturing industry, according totheir level of creative destruction. Here, creative destruction is measured byhalf the sum of job creation and job destruction rates from the QuarterlyWorkforce Indicators series from the Census (See Davis et al. [1996]); (ii) forany value x on the horizontal axis, we group all the sector-years with level xon the creative destruction scale, and compute the corresponding within-group average correlation y � x � between current yearly TFP growth and thecurrent yearly number of patents.

Philippe Aghion —————————————————————————————————————————————————————— 703


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

Then we see that the correlation between TFP growth and patenting overtime is lower in sectors with higher rates of creative destruction.

Third and last: my optimism regarding future growth prospects is alsobased on the observation that many countries have taken only belated andincomplete advantage of technological advances, e.g. because of structuralrigidities or inappropriate economic policies. These countries did not adaptand therefore did not evolve from a catch-up economy to an innovationeconomy. The comparison between Sweden and Japan is highly informa-tive: growth of productivity accelerated in Sweden as it declined in Japan.

While it is hard to talk about *stagnation* in the US and in other devel-oped economies once we have added the missing growth from imputation,yet one needs to explain why TFP growth increased in the US between themid-1990s and the mid-2000s, before declining thereafter. And we must lookfor an explanation that can be reconciled with several empirical facts, inparticular: (i) an increase in the average mark-up within sectors, which isentirely happening through between-firm reallocation and not within firms;(ii) a declining rate of firm entry; (iii) a declining investment; (iv) an increasedmarket concentration and a declining labor share, both of which are drivenby between-firm reallocation. Aghion et al. [2018b] develop a story based onthe emergence of super-star firms during the 1995-2005 period, which inturn has to do with the IT revolution. Super-star firms invade a whole rangeof new and/or existing product lines where they deter innovation, first andforemost by low-productivity innovators (firms with lower mark ups) and to

Fig. 5 Creative destruction and the correlation between TFP

growth and patenting



© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

a lesser extent by high-productivity innovators (firms with higher mark ups).This in turn explains the slowdown in firm entry and in (R&D) investmentfollowing the IT wave, and it also explains why the increase in averagemark-up and average market concentration, is mainly due to high-productivity innovators prevailing over low-productivity innovators.

The above discussion has potentially important implications for how AI mayaffect productivity growth. On the one hand, AI automates not only tasks in theproduction of goods and services, but also tasks in the production of ideas.This in turn should help accelerate productivity growth, and achieve infiniteincome in finite time (we call that a *singularity*, see Aghion, Jones andJones [2018]. However, a lack of appropriate competition institutions maylead to the opposite outcome, for example by allowing for new super-starfirms to emerge and stall the entry of subsequent innovative firms.

4.3. Innovation, inequality, and social mobility

Over recent decades, developed nations have experienced an acceleratedincrease in income inequality, especially at the top tier, with the top 1%capturing a rapidly growing share of total income (Atkinson et al. [2011];Piketty [2013]). What explains this evolution?

Figure 6 compares the evolution of innovation in the United States since1960 (as measured by the number of patents registered annually with theUnited States Patent and Trademark Office), with extreme inequality (as mea-sured by the share of income attributed to the top 1% of earners). The similar-ity in the two curves (innovation and the top 1%’s share of income) is striking.

Fig. 6 The top 1% income share and the annual patent flow

in the US since 1960

Philippe Aghion —————————————————————————————————————————————————————— 705


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

A new study by Antonin Bergeaud, Richard Blundell, Ufuk Akcigit, DavidHemous, and myself (Aghion et al. [2015]) shows that this strong correlationreflects a causal link between innovation and extreme inequality: incomefrom innovation contributes significantly to the increase in the share ofincome going to the top 1%.

The observation that the observed increase in the top 1% results in partfrom innovation, and not solely from returns from real estate and specula-tion, provides an important insight, because innovation has virtues that theother sources of high income do not necessarily share.

First, as previously mentioned, innovation is the main motor of growth indeveloped economies. Second, although in the short term innovation ben-efits those who generated or enabled the innovation, in the long term itsreturns are dissipated due to imitation and creative destruction. In otherwords, the inequality induced about by innovation is temporary. Third,because of the link between innovation and creative destruction, innovationgenerates social mobility: it enables new talent to enter the market and todisplace (partially or totally) the firms in place. Thus in the United States,California —currently the most innovating state of the union— far outpacesAlabama —which is among the least innovating states— both in terms of theinequality of income going to the top 1% and in terms of social mobility.

Two figures are especially eloquent. The first figure (Figure 7) describesthe relationship between innovation and social mobility by comparingAmerican municipalities. Social mobility is defined as the probability that anindividual from a modest background (i.e., one whose parents were in thelowest quintile in the earnings scale between 1996 and 2000) will reach thehighest quintile in 2010 upon reaching adulthood (based on the work ofChetty et al. [2014]). Innovation is measured by the number of patents filedwith the United States Patent and Trademark Office per resident in themunicipality. The resulting graph shows a strong positive correlationbetween innovation and social mobility.

Fig. 7 Innovation and social mobility



© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

The second figure (Figure 8) shows that there is no correlation betweeninnovation and the broader measures of inequality such as the Gini coeffi-cient, which measures the deviation between the actual distribution ofincome within of an economy and a perfectly equal distribution.

By taking into account all of the pieces of the puzzle, we can respond to thequestion of whether we should object to innovation on the grounds that it con-tributes to income inequality. The response is no, because innovation generatesgrowth. It does not increase inequality in broader terms; rather it stimulates so-cial mobility. As a corollary to this discussion, tax policy must differentiate be-tween innovation and other sources of top income. Put differently, we must dis-tinguish between a Steve Jobs and a Carlos Slim. Tax policy that discouragesinnovation would not only inhibit growth, but would also reduce social mobility,whereas innovation does not increase inequality measured broadly.

4.4. Firm dynamics

Again using both, theory and empirical analysis, younger generations ofscholars have developed the Schumpeterian paradigm to look at the impor-tant question of the relationship between growth, innovation and firmdynamics. Here I should mention Daron Acemoglu, Ufuk Akcigit, NickBloom, Chang Tai Hsieh, William Kerr, Pete Klenow, Tor Jakob Klette andSamuel Kortum as prominent figures in this new development.

The empirical literature on firm size distribution and on firm dynamics,had pointed to a number of interesting stylized facts. In particular: (i) the firmsize distribution is highly skewed, as shown in Figure 9; (ii) firm size and firmage are highly correlated; (iii) small firms exit more frequently, but the onesthat survive tend to grow faster than the average growth rate.

Fig. 8 Innovation percentiles, and the top 1% and Gini99

inequality measures

Philippe Aghion —————————————————————————————————————————————————————— 707


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

Incidentally, these are all facts that non-Schumpeterian growth modelscould not explain. Indeed, the first four facts required a new firm to enter,expand, and then shrink over time, and eventually be replaced by newentrants: these and the last fact on the importance of reallocation are allassociated with the Schumpeterian idea of creative destruction.

An important milestone to try and explain the above facts and more gen-erally the relationship between growth, innovation and firm dynamics, wasthe Schumpeterian model by Klette and Kortum [2004]. In this model, firmsare defined as a collection of production units where successful innovationsby incumbents will allow them to expand in product space (see Figure 10).

Fig. 9 Firm Size Distribution with Multiproduct Firms

Fig. 10 The firm as a collection of product lines



© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

Let us show why this paradigm allows us to explain the above stylizedfacts. Consider first the prediction that the size distribution of firms is highlyskewed. Recall that in this model, firm size is summarized by the number ofproduct lines of a firm. Hence, a firm needs to have succeeded manyattempts to innovate in new lines and at the same survived many attemptsby potential entrants and other incumbents at taking over its existing lines,in order to become a large firm. This is turn explains why there are so fewvery large firms in steady-state equilibrium, i.e. why firm size distribution ishighly skewed as shown in a vast empirical literature.

Consider now the prediction that firm size and firm age are positivelycorrelated. In the Klette-Kortum model, firms are born with a size of 1.Subsequent successes are required for firms to grow in size, which naturallyproduces a positive correlation between size and age. This regularity hasbeen documented extensively in the literature.

Now coming to the prediction that small firms exit more frequently, andthat the ones that survive tend to grow faster than average: in the Klette-Kortum model, it takes only one successful entry to make a one-product firmto exit, whereas it takes two successful innovations by potential entrants tomake a two-product firm exit. The facts that small firms exit more frequentlyand grow faster conditional on survival have been widely documented in theliterature.

5. Conclusion

In this lecture we have argued that models where growth is entirely basedon capital accumulation miss important aspects of the growth process, andthat the Schumpeterian growth paradigm sheds light on several enigmasassociated with growth: in particular, the relationship between competitionand growth, the transition trap, the debate on secular stagnation, the rela-tionship between growth and inequality, and the relationship betweengrowth and firm dynamics.

The Schumpeterian growth paradigm, combined with empirical analysisusing micro data, can be used to explore many more questions. Below wejust provide a few examples of such questions.

One line of research is to look at how firm dynamics, and thereforeinnovation-based growth, are affected by institutions: for example, labormarket institutions, the democratic or non-democratic nature of politicalinstitutions, the scope for corruption and the strength of counter-powers tolimit it, etc.

A second line of research is to revisit growth policy design through thelens of firm dynamics and innovation: tax policy, patent policy, competitionpolicy, the financing of electoral campaigns, labor market policies, tradepolicy, education, etc.

A third line of research is to look at how human capital and innovationinteract with one another: in particular, when does creative destruction and

Philippe Aghion —————————————————————————————————————————————————————— 709


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

turnover contributes positively or negatively to human capital accumula-tion?

A fourth line of research is to use individual data to identify the charac-teristics of innovators and how the returns to innovation are shared withco-employees and the entrepreneur within the same firm or in the economy.

A fifth line of research is to explore the growth measurement issuedeeper: i.e., to identify additional sources of growth mismeasurementbeyond imputation, and to evaluate how much additional productivitygrowth is missed.

Finally, our emphasis on innovation-based growth does not mean that weshould disregard capital accumulation and the role of savings, quite thecontrary. There has been an important literature looking at capital embodiedtechnical progress, and we devoted whole chapters to capital accumulationand Schumpeterian innovation with Peter Howitt in each of our two books(Aghion and Howitt [1998], [2009]).

More recently, important papers have looked at how major innovationshave both, the effect of replacing labor by capital in existing production lines(this is referred to as automation) and of creating new lines (e.g. see Aghion,Jones and Jones for references [2018]). This in turn suggests very fruitfulavenues for further research on growth and innovation.

References

ACEMOGLU D., AGHION P., ZILIBOTTI F. [2006], Distance to Frontier, Selection, andEconomic Growth, Journal of the European Economic Association, 4(1), 37-74.

AGHION P., HOWITT P. [1992], A Model of Growth Through Creative Destruction,Econometrica, 60(2), 323-351.

AGHION P., HOWITT P. [1998], Endogenous Growth Theory, Cambridge, MA: MITPress.

AGHION P., HOWITT P. [2009], The Economics of Growth, Cambridge, MA: MITPress.

AGHION P., HARRIS C., VICKERS J. [1997], Competition and Growth with Step-by-Step Innovation: An Example, European Economic Review, Papers and Proceed-ings, 41(3-5), 771-782.

AGHION P., HOWITT P., PRANTL S. [2013], Patent Rights, Product Market Reformsand Innovation. NBER Working Paper Series n° 18854.

AGHION P., JONES B., JONES C. [2018], AI and Economic Growth, Mimeo Stanford.AGHION P., AKCIGIT U., BERGEAUD A., BLUNDELL R., HEMOUS D. [2015], Innova-

tion and Top Income Inequality, NBER Working Papers Series n° 21247.AGHION P., BERGEAUD A., BOPPART T., KLENOW P., LI H. [2017], Missing Growth

from Creative Destruction, Mimeo London School of Economics.AGHION P., BLOOM N., BLUNDELL R., GRIFFITH R., HOWITT P. [2005], Competition

and Innovation: An Inverted-U Relationship, Quarterly Journal of Economics,120(2), 701-728.



© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

AGHION P., BLUNDELL R., GRIFFITH R., HOWITT P., PRANT S. [2009a], The Effects ofEntry on Incumbent Innovation and Productivity, Review of Economics and Sta-tistics, 91(1), 20-32.

AGHION P., BOUSTAN L., HOXBY C., VANDENBUSSCHE J. [2009b], The CausalImpact of Education on Economic Growth: Evidence from the U.S., Mimeo Har-vard.

AGHION P., HARRIS C., HOWITT P., VICKERS J. [2001], Competition, Imitation andGrowth with Step-by-Step Innovation. Review of Economic Studies, 68(3), 467-492.

AKCIGIT U., ALP H., PETERS M. [2014], Lack of Selection and Limits to Delegation:Firm Dynamics in Developing Countries, Mimeo University of Pennsylvania.

ATKINSON A., PIKETTY T., SAEZ E. [2011], Top Incomes in the Long-Run History,Journal of Economic Literature, 49(1), 3-71.

BLUNDELL R., GRIFFITH R., VAN REENEN J. [1995], Dynamic Count Data Models ofTechnological Innovation. Economic Journal, 105(429), 333-344.

BLUNDELL R., GRIFFITH R., VAN REENEN J. [1999], Market Share, Market Value andInnovation in a Panel of British Manufacturing Firms, Review of Economic Stud-ies, 66(3), 529-554.

BOLDRIN M., LEVINE D. [2008], Against Intellectual Monopoly, Cambridge, MA:Cambridge University Press.

CHETTY R., HENDREN N., KLINE P., SAEZ E. [2014], Where Is the Land of Opportu-nity? The Geography of Intergenerational Mobility in the United States, QuarterlyJournal of Economics, 129(4), 1553-1623.

DAVIS S., HALTIWANGER J., SCHUH S. [1996], Job Creation and Destruction, Cam-bridge, MA: MIT Press.

GORDON R. [2012], Is U.S. Economic Growth Over? Faltering Innovation Confrontsthe Six Headwinds, NBER Working Papers Series n° 18315.

HANSEN A. [1939], Economic Progress and Declining Population Growth. AmericanEconomic Review, 29(1), 1-15.

KLETTE T., KORTUM S. [2004], Innovating Firms and Aggregate Innovation, Journalof Political Economy, 112, 986-1018.

MALINVAUD E. [1953], Capital Accumulation and Efficient Allocation of Resources,Econometrica, 21, 233-268.

PIKETTY T. [2013], Le Capital au XXe Siècle, Paris, Seuil.ROMER P. [1990], Endogenous Technical Change, Journal of Political Economy,

98(5), 71-102.SCHERER F. [1967], Market Structure and the Employment of Scientists and Engi-

neers, American Economic Review, 57(3), 524-531.SUMMERS L. [2013], Why Stagnation Might Prove to Be the New Normal, Financial

Times, December 15, 2013.

Philippe Aghion —————————————————————————————————————————————————————— 711


© D

allo

z | T

éléc

harg

é le

26/

08/2

022

sur

ww

w.c

airn

.info

(IP

: 65.

21.2

29.8

4)©

Dalloz | T


w.cairn.info (IP

: 65.21.229.84)

Innovation and Growth from a Schumpeterian Perspective

Documents

Transcript of Innovation and Growth from a Schumpeterian Perspective