Economic crisis and innovation

Andrea FilippettiLondon School of Economics and Political Science, and

Italian National Research Council, CNR

Banca d'Italia lunedì 25 novembre

Creative destruction or creative accumulation?

Schumpeter 1911: economic environment characterized by fierce competition in which the new firm is the driver of innovation activity supported by the creation of new credit by the banking system

Schumpeter 1942:emphasized the role of well established large firms;market structure has shifted into oligopolistic competition

Gerhard O. Mensch

1979: Stalemate in Technology

stagnation and innovation for recovery overlap: “at the same time that wide areas of current economic interests are gripped by stagnation, creative progress is building in new areas of activity […] in the technological stalemate, the economy becomes structurally ready for basic innovations”

The SPRU response: It is innovation diffusion that matters!

Freeman, Clark and Soete (1982)

In a Schumpeterian vein, there are winners and losers

• And the existence of losers is the guarantee that there will be some winners

• Nations, industries and companies that will manage to introduce and adapt to the technological change are likely to grow and prosper

• Those that will not be able to do so, will be most affected by the economic crisis

• Implicit assumption: barriers to enter new technological fields are low

A dissenting voice: Keith Pavitt

• Competences are generated in a cumulative pattern

• This applies for international patterns of technological accumulation

• It also applies at to company level• Little possibility for new entrants when facing

large innovative firms

Technological Accumulation

To innovate, companies need to develop competences, which are created through experience. Innovators of the future are likely to be the innovators of the past

Persistence is the key factor in generating successful innovations

Also nations generally continue to innovate in the areas where they are traditionally strong

Management literature…

Are creative destruction and technological accumulation sensitive to the business cycle?

• During economic expansion, innovative firms lead technological change also by increasing their investment in innovation (supporting technological accumulation)

• Economic crises generate turbulence and some new entrants are willing to spend more to innovate, also in blue sky explorations (creative destruction). Resources made available can be used for the purpose

2 levels of analysis

• Firm level

• Country level

An application with micro data on panel analysis

Economic crisis and innovation: Is destruction prevailing over accumulation?

Daniele Archibugi, Andrea Filippetti, Marion Frenz

http://www.sciencedirect.com/science/article/pii/S004873331200162X

Our argument• there is a general consensus on the fact that the most

innovative firms are also more likely to persist in innovating

• firms with a more agile/flexible structure might take better advantage of changing environments and new market opportunities

• the unique environment of the current economic crisis might challenge innovation in a cumulative fashion and lead to an environment more closely related to creative destruction

4 hypothesesHypothesis 1. During a crisis innovation investment concentrates further in those firms that were already highly innovative before the crisis

Hypothesis 2.Increased investment in innovation during the crisis is more strongly correlated with two groups of firms – (a) those previously classified as great innovators and (b) those classified as fast growing new entrants

Hypothesis 3Increase in investment in innovation before and during the crisis is positively associated with internal R&D, firm size and firm internal financial resources

Hypothesis 4.Firms that follow mixed strategies of exploitation and exploration – ambidextrous firms – are more likely to increase investment in innovation during the crisis compared with before

Data: The UK Innovation Surveys

• under 2,500 enterprises that responded to the latest three waves of the UK version of the CIS (2004, 2006 and 2008)

• we analyse a balanced panel with observations at three points in time (T=3).

The dependent variables

Variables N. of firms MeanTotal innovation expenditure per employee in 2006 in £000s 2,479 2.44

Total innovation expenditure per employee in 2008 in £000s 2,485 2.04

we require a measure of the change in innovation related investment before and during the crisis

DV1. We compute the change in 2008 compared with 2006 and use this as the change in innovation expenditure during the crisis.

DV2. Before the crisis is the change in innovation investment in 2006 compared with 2004

Variable Name Description Hypothesis1 Log change in

innovation expenditure in 2006 and 2008

Log of innovation related investment compared to previous period

Dependent variable

2 Log total innovation expenditure in 2004 and 2006

Log of innovation expenditure in the previous period

Control variable

3 Great innovators in 2004

Dummy variable. Great innovators are enterprises that introduced new-to-the-market goods and services in 2004

Testing H1 and H2 - Great innovators increase innovation expenditure during the crisis

4 Newly established 2000

Dummy variable. Enterprises established between 2000 and 2004, value 1, others 0

Control variable

5 Growth of newly established firms in 2006 and 2008

Log of the change in turnover compared to previous period for new firms as defined in (4). This variable takes a value of zero for firms established before 2000

Testing H2 – Fast growing new enterprises increase innovation expenditure during the crisis

6 Internal R&D in 2004 and 2006

Dummy variable. Enterprises with internal R&D expenditure in the previous period, value 1, others 0

Testing H3 – Enterprises with internal R&D increase innovation expenditure during the crisis

7 Log employees in 2004 and 2006

Size of the firm according to the number of employees in the previous period

Testing H3 – Large enterprises increase innovation expenditure during the crisis

8 Availability of finance in 2004 and 2006

Dummy variable. Firms which gave in the previous period medium or high importance to the availability of finance as innovation obstacle, value 1, firms that gave no or low importance, value 0

Testing H3 – Enterprises with internal financial resources increase innovation expenditure during the crisis

9 Log sales per employee in 2004 and 2006

Log of sales per employee in the previous period

Testing H3 – Enterprises with higher sales per employee (as proxy of available internal resources) increase innovation expenditure during the crisis

10 Exploration in 2006 and 2008

Dummy variable. Firms in the upper two quartiles in the sum of the scores across four-point likert scales in the question: “how important were each of the following factors in your decision to innovate: (i) increase range of goods or services; (ii) entering new markets or increased market share”, value 1, others 0.

Control variable

11Exploitation in 2006 and 2008

Dummy variable. Firms in the upper to quartiles in the sum of the scores across four-point likert scales in the question: “ how important were each of the following factors in your decision to innovate: (i) improving quality of goods or services; (ii) improving flexibility for producing goods or services; (iii) increasing capacity for producing goods or services; (iv) reducing costs per unit produced

Control variable

12 Ambidexterity in 2006 and 2008

Dummy variable. A firm is in the upper quartiles with respect to both - exploration and exploitation (see 11 and 12), value 1, others 0

Testing H4 – Enterprises that follow mixed strategies of exploitation and exploration – ambidextrous enterprises, increase innovation expenditure during the crisis

13 IPRs in 2004 and 2006

Dummy variable. Firms that declared to use IPR protection in the previous period, value 1, others 0

Control variable

14 Skills in 2006 and 2008

Log of the proportion of employees that hold a degree at BA/BSc level or above.

Control variable

15 International markets in 2006 and 2008

Dummy variable. Enterprises that operate outside the UK, value 1, others 0

Control variable

n. of firms

Percentage

Share of innovation

exp. 2006

Share of innovation

exp. 2008

Average innovation

exp. 2006 in £000s

Average innovatio

n exp. 2008 in

£000s

Change in average

innovation exp. 2006-

2008

All other firms 2,161 87 0.79 0.63 563 413 -0.27

Great innovators 324 13 0.21 0.37 981 1,599 0.63

Total 2,485 100 1.00 1.00 618 568 -0.08

Innovation expenditure of great innovators and other firms, 2006 and 2008

Hypothesis 1. During a crisis innovation investment concentrates further in those firms that were already highly innovative before the crisis

before the crisis during the crisis

Great Innovators 2004 not significant + (**)

Newly established 2000 - (***) - (*)

Fast grow. new firms t not significant + (**)

In-house R&D t-1 not significant + (**)

Log employees t-1 + (**) + (**)

Availability finance t-1 not significant not significant

Sales per employee t-1 + (**) + (**)

Explorative strategy t 0.31* 0.6**

Exploitative strategy t 0.6** 0.38*

Regression resultsInnovation behaviour before and during the crisis. DV: change in innovation expenditure before and during the crisis

Conclusions:• Firms in our sample reduce innovation

expenditure in 2008 by 8 percent compared to 2006. No doubt that the crisis has brought some “destruction”

• strong support for creative accumulation. Firms identified as the great innovators in 2004 are responsible for a larger share of innovation expenditure in 2008

• being a great innovator does not predict increase in innovation investment before the crisis, but it does during the crisis

... continued

• another category of firms which is gaining momentum during the crisis: they are the fast growing new firms

• this group of firms does not show an above average behaviour in 2006 but it starts to increase expenditure during the crisis.

... continued

• size and economic performance play a less important role.

• presence of in-house R&D activity becomes a major predictor of increase in innovation expenditure during the crisis

• firm’s strategy, pursuing an explorative strategy (including looking into new markets), becomes relatively more important.

INNOVATION PERSISTENCE AND THE INSTITUTIONAL SETTING

24

National Innovation System: the founding fathers

Freeman, Lundvall, Nelson

25

National systems of innovation

Firms

Government

Science & technology institutions;

e.g. R&D labs, unis

Educationinstitutions

Industrye.g., competitors,

suppliers

Financial institutionse.g. banks,

venture capitalists

Source: Smith (2006), p. 295

do specific configurations of NSI provide a comparative institutional advantage in times of crisis?

Thus the first general question at stake is whether structure matters vis-à-vis demand during a big recession

Data and methodology (1)

Innobarometer 2009 (European Commission) – firm level survey on more than 5000 firms across Europe – April 2009

Question no. 1: “Compared to 2006, has the amount spent by your firm on all innovation activities in 2008 increased, decreased, or stayed approximately the same?”

Question no. 2 “In the last six months [November 2008 to April 2009] has your company taken one of the following actions [increased, decreased or maintain the innovation spending] as a direct result of the economic downturn?”

Data and methodology (2)

European Innovation Scoreboard 2008 (EIS)(European Commission) aims at measuring and comparing the

innovation performance at country level based on a composite indicator:

The EIS Summary Innovation Index: composed of 29 variables addressing 7 dimension of a country’ system of innovation

Firms’ innovation investments: comparison between the three years period 2006-2008 and the first six months of 2009

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

Increassed%

Decreased%

Stayed thesame %

Per

cent

ages

of

firm

s w

hich

are

rep

orte

d to

hav

e in

crea

sed,

de

crea

sed

or m

aint

aine

d th

eir

inno

vatio

n ex

pend

iture

s

Firms' innovation expenditures 2006-2008 Firms' innovation expenditures 2009

Is the crisis impairing the catching-up process of innovation capabilities?

Austria

Belgium

Bulgaria

Czech rep.

Denmark

Estonia

Finland

France

Germany

Greece

Ireland

Latvia

Lithuania

Luxemburg

Netherlands

Norway

Poland

Portugal

Romania

Slovakia

Slovenia

Sweden

United Kingdom

HungaryItaly

Spain

Switzerland

0.2

0.3

0.3

0.4

0.4

0.5

0.5

0.6

0.6

0.7

0.7

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

InnoInv 09 performance

Inno

Str

uct

perf

orm

ance

Innovation investments balance = 0 in 2009

The impact of the current recession on firms’ innovation investments

-90

-80

-70

-60

-50

-40

-30

-20

-10

0

Sw

itzer

land

Den

mar

k

Net

herl

ands

Fin

land

Aus

tria

Luxe

mbu

rg

Spa

in

Bel

gium

Uni

ted

Kin

.

Por

tuga

l

Irel

and

Est

onia

Ital

y

Ger

man

y

Hun

gary

Cze

ch r

ep.

Nor

way

EU

27

Slo

veni

a

Sw

eden

Latv

ia

Fra

nce

Slo

vaki

a

Bul

gari

a

Pol

and

Rom

ania

Gre

ece

Lith

uani

a

Diff

eren

ce b

etw

een

the

firm

s’ in

vest

men

ts b

alan

ce

in t

he m

ediu

m a

nd s

hort

-ter

m

Change in the behaviour of the firm related to its innovation investment as a response to the crisis vis-à-vis the period before

the crisis

Variable Value assumed by the variable

Behaviour of the firm

Change in innovation behaviour (INVchange) = [- (INVEST2 – INVEST1)]

= 1

Cyclical (e.g. firms which were increasing and pass to maintaining or decreasing in response to the crisis)

= 0

Neutral (e.g. firms which were increasing and keep on increasing)

= -1 Counter-cyclical (e.g. firms which were decreasing and pass to increasing or maintain)

Operationalizing the national innovation system

• point out those country-specific features which have a role in offsetting the cyclical behaviour of the firms, and therefore that have a positive influence on persistency of innovation investment.

• The following different characteristics of the NSI have been derived from the EIS: – i. the stock of knowledge; – ii. the quality of the human resources; I– ii. the depth of the financial and credit system; – iv. the specialization of the country

NIS characteristics Variable Indicator

Stock of knowledge

Business R&D Business R&D expenditures (% of GDP)

Public R&D Public R&D expenditures (% of GDP)

Non-R&D expenditure Non-R&D innovation expenditures (% of turnover)

EPO patents EPO patents per million population

IT expenditures IT expenditures (% of GDP)

Human resources

S&E and SSH graduates

S&E and SSH graduates per 1000 population aged 20- 29 (first stage of tertiary education)

S&E and SSH doctorate graduates

S&E and SSH doctorate graduates per 1000 population aged 25-34 (second stage of tertiary education)

Tertiary education Population with tertiary education per 100 population aged 25-64

Life-long learning Participation in life-long learning per 100 population aged 25-64

Youth education Youth education attainment Level

Credit system Venture capital

Venture capital (% of GDP)

Private credit Private credit (% of GDP)

Industrial Specialization

Employment in medium-high & high-tech manufacturing

Employment in medium-high & high-tech manufacturing (% of work-force)

Employment in knowledge-intensive services Employment in knowledge intensive services (% of workforce)

Operationalizing short-term conditions

-0.10

-0.05

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

Est

onia

Lat

via

Lit

huan

ia

Pola

nd

Rom

ania

Hun

gary

Swed

en

Irel

and

UK

Slov

enia

Nor

way

Mal

ta

Spai

n

Cze

ch R

.

Bul

gari

a

Den

mar

k

Finl

and

Port

ugal

Slov

akia

Bel

gium

Ital

y

Net

herl

ands

Aus

tria

Fran

ce

Ger

man

y

Lux

embo

urg

Gre

ece

Swit

zerl

and

domestic demand drop export drop

The drops in the domestic demand and export, 1st term 2008 – 3rd term 2009

Model no. 1 Model no. 2 Model no. 3

Demand effects

Domestic demand drop 1.26*** 0.52 0.72

Export drop 1.72*** -0.60 -0.43

National Innovation System effects

Knowledge 0.07 1.82

Human Resources -0.13 4.77**

Venture capital -0.02 -0.15

Private credit -0.41*** -0.54***

High-tech manufacturing specialization -0.73*** -3.33***

Knowledge intensive service specialization 0.29 -2.25**

Interaction effects

Demand*knowledge 0.19

Demand*human resources -4.81**

Demand*high-tech specialization 1.35

Demand* KIS service specialization 2.45**

Export*knowledge -1.57

Export*human resources -4.75**

Export*high-tech specialization 2.72***

Export* KIS service specialization 3.46**

Firm level control variables

Medium and large firms 0.24*** 0.22*** 0.22***

Highly innovative firms 0.78*** 0.79*** 0.82***

Internationalised firms 0.06 0.13 0.13

Industry dummies included included included

Observations 3072 3072 3072 Robust standard errors in parentheses (country clustered errors provide the same results) *** p<0.01, ** p<0.05, * p<0.1 Reference control variables: small firms; low innovative firms.

Ordered logit model, robust estimates (dependent variable: INVchange)

Is innovation cyclical or persistent?• One of the most significant results of our analysis is

that about sixty-five per cent of the firms declare to have kept their innovation investment unchanged in spite of the crisis

• This somehow confirms the importance of:– technological accumulation (stressed, among others, by

Nelson and Winter, 1982; Grandstrand et al., 1997; Patel and Pavitt, 1997),

– and lends substantial support to the persistency of innovative activities over time (Geroski et al., 1997; Cefis and Orsenigo, 2001).

The uneven effects of the crisis and the role of National Systems of Innovation

• the crisis has not been of the same magnitude across all European countries. On the contrary, we have shown that the most negatively affected by the downturn are those EU New Member States which were catching up over the 2006-2008 period

• We have attempted to explain this evidence on the ground of the structural characteristics of the NSI and the role played by domestic demand and export.

The structural characteristics of the NSI seem to play a more relevant role than demand.

The role of the institutional framework

Innovation, labour market institutions and skills

Social protection and the formation of skills

Labour market institutions

• different institutional arrangements have been identified namely along three main dimensions: – (i.) unemployment security, – (ii.) employment protection, and – (iii.) vocational and educational training (VET).

Labour attachment• the attachment of factors of production to a firm

favors incremental innovation and discourages radical innovation

• CMEs specialize in incremental innovation because employment protection, low inter-firm mobility, insider control of firms and weak markets for corporate control

• LMEs are said to specialize in radical innovation because the factors of production can be rapidly mobilized for a promising new project, and de-mobilized if it fails

2 types of attachment mechanisms

• Institutional constraints: i.e. EP - that make it costly for the firm to shed labor; this type of exogenous attachment encourages investment in specific skill and generates ongoing teams of skilled production workers (exogenous)

• attachment is not imposed institutionally through EP, but created endogenously by skill. A worker with experience in a particular firm, or a team of such workers, may generate for a firm a quasi-rent – a situation in which the worker is more valuable when employed by the firm than in the worker’s next-best employment

Absolute values

CountryEmployment protection

index(EP)

Replacement rate(RR)

Vocational and education training

(VT)Austria 1.93 62.21 77.05Belgium 2.18 63.14 55.76Czech Republic 1.96 67.36 73.72Denmark 1.5 77.76 53.74Finland 1.96 72.4 54.79France 3.05 71.38 43.8Germany 2.12 66.46 57.48Greece 2.73 49.14 32.06Hungary 1.65 73.22 26.64Ireland 1.11 59.62 32.67Italy 1.89 65.52 59.77Luxembourg 3.25 87.64 60.37Netherlands 1.95 77.95 67.64Norway 2.69 72.22 55.24Poland 1.9 64.14 46.54Portugal 3.15 78.91 33.24Slovak Republic 1.44 59.78 71.88Spain 2.98 72.83 43.83Sweden 1.87 70.52 59.46Switzerland 1.14 79.65 64.8United Kingdom 0.75 57.05 23.61

Country-level variables

Dependent variable: innovation persistence 2009

(1) No

Interactions (2) With

Interactions (3) Sample selection

Firm level control variables

Innov. Invest Down 2006-8 -0.877*** -0.867*** -0.810***

-0.0857 -0.0842 -0.0802 Innov. Invest Up 2006-8 0.0189 0.0177 0.0648

-0.076 -0.0774 -0.0713

Sharp Turnover Decline Pre-Crisis -0.645*** -0.641*** -0.574***

-0.13 -0.131 -0.137 Some Turnover Decline Pre-Crisis -0.310*** -0.307*** -0.327***

-0.0777 -0.0791 -0.0877

Country level variables

VET (vocational educational and training) 0.317 -1.952*** -2.122***

-0.176 -0.539 -0.483

RR (unemployment protection) 0.385* -0.633 -0.816*

-0.159 -0.371 -0.408

EP (employment protection) 0.0715 -0.535* -0.505*

-0.122 -0.216 -0.225

VET*RR

3.961*** 4.427***

-1.058 -1.028

VET*EP 4.209*** 4.282***

-1.091 -0.984

RR*EP

1.488*** 1.614***

-0.443 -0.47

VET*RR*EP -6.744*** -7.126***

-1.608 -1.493

Country level control variables

Delta GDP to q1 2009 -0.022 -0.0507*** -0.0475***

-0.0129 -0.0128 -0.0119 Delta GDP to q1 2008 -0.0175 0.0229 0.0269 Constant 0.285 0.433** 0.367* -0.149 -0.168 -0.181 SELECTION MODEL

Firm size: 50-249 employees 0.304***

-0.0685 250-499 employees 0.404***

-0.0996

500 + employees

0.624***

-0.0824

Internationalized firm 0.333***

-0.0571

Constant

0.822***

-0.08

Observations 3473 3473 3608 Pseudo R2 0.072 0.076

AIC 3589.2 3577 5883

conclusions

• RR (our measure of unemployment protection) and VET are complementary.

• So are EP and VET. • When both forms of income insurance are

low, VET is actually associated with lower levels of innovation persistence

• income insurance / VET systems have a comparative advantage in the production of specific skills, this supports the view that specific skills are better for sustaining innovation during a crisis than general skills.

• We find no substantial difference between the

performance of flexicurity (high RR, high VET) and non-liberal (strong EP, high VET) systems.