Disentagling the relations_between_learning_modes

Disentangling the relationship

between learning modes, innovation

output and economic performance:

Evidence from Spain

González-Pernía, J.L.

Parrilli, M.D.

Peña-Legazkue, I.

‘Regional Innovation in the Global Economy’ Institute for Research on Innovation (IRIS)

Stavanger, 11 December 2012

Motivation

• How does innovation transform into economic value? There is a need for a better understanding on how learning (for innovation) is managed within organizations.

• Is there any relationship between different learning approaches (STI and DUI), innovation output (product and process innovation) and business performance (sales growth)?

• Is it as expected on the basis of former studies (Jensen et al., 2007; Isaksen and Karlsen, 2011; Parrilli & Elola, 2011)?

Key Concepts

• STI learning mode: learning and innovation based on science and technology drivers such as R&D expenditure, human capital in S&T disciplines, investment in infrastructures.

• DUI learning mode: learning and innovation based on learning-by-doing, by-using and by-interacting, thus based on interactive experience and practice.

Literature on “STI-DUI”

• Traditional approach to Innovation based on Science and Technology drivers (STI approach): – R&D expenditure and infrastructure (Griliches, 1979; Greunz,

2005)

– human capital (Romer, 1994)

• This approach tends to identify determinants of: – innovation outputs (patents)

– economic performance (productivity growth)

• However, this approach does not explain the capacity of some economies (Denmark, Norway, Italy, …) to innovate and to be competitive despite their more modest effort in STI activities (i.e., reduced investment in R&D).

Literature on “STI-DUI”

• On these bases, Lundvall and the Scandinavian school

identified other drivers of innovation: later known as

innovation based on learning-by-Doing, by-Using and by-

Interacting (DUI) (Lundvall, 1992; Lundvall and Johnson,

1994).

• Interaction among all workers and firms, and daily practice

help increase innovation and have an impact on economic

performance.

Literature on “STI-DUI”

• Jensen, Johnson, Lorenz and Lundvall (2007) tested the relevance of both innovation modes, STI and DUI, on 4000 Danish firms (DISKO) concluding that the most productive innovation mode is the combination of the two: STI+DUI.

• Two recent studies (Aslesen et al., 2011 on Norway, and Chen et al., 2011, on China) confirm the result. Another one offers more nuanced results, i.e. international DUI vis-a-vis local DUI (Dahl-Fitjar and Rodríguez-Pose, 2012)

• & Parrilli and Elola (2011) found that such combination may not be the most effective in all contexts.

Contribution

What we know …

Drivers Outcome

STI

Innovation (innovation black box!)

DUI

What we don´t know … (though Dahl-F. & Rodriguez-P. worked on this)

Drivers Modes Performance

STI

Product Innovation Sales

Process Innovation Growth

DUI

Contribution (cont.)

What we seem to know …

• Combined STI-DUI approach leads to best new-to-market innovation results (Jensen et al., 2007)

What we don´t know …

• The STI and DUI effect (by considering both intra-firm and inter-firm indicators) on product and process innovation.

• Is STI more tighly linked to product innovation and DUI more tighly linked to process innovation or viceversa?

Contribution (cont.)

What we know …

• Both product and process innovation increase firm performance under a static view

• Most studies: one-country, one-period

What we don´t know …

• Whether this relationship works under a dynamic view

• Results for mutliple-regions and multiple-periods

Proposed Framework

Sales Growth

Product innovation

Process innovation

STI-mode of learning

DUI-mode of learning

STIxDUI-mode of learning

Strong relationship Weak relationship

DRIVERS MODES PERFORMANCE

Propositions

• Although, the STI-DUI combination is likely to be the most effective approach, it needs to be further studied.

• More specifically, our propositions are:

P1: STI higher effect on product innovation

DUI higher effect on process innovation

STI*DUI higher effect on both product/process

P2: Product innovation stronger effect on late performance

Process innovation stronger effect on early performance

Data and methodology • Source: Technological Innovation Panel (PITEC)

• Period of analysis: 2004-2009

• Unit of analysis: Firms

– 10 or more employees

– No incidence in employment

– Manufacturing and services

Nº cases: 8,682 firms

Average period: 3.9 years

Overall sample: 33,789 obs.

Basic empirical model: Random effects

)(= ,,1-ti,ti, tiiti uZXY

Description of variables

Construct Variables Measurement

Factor loadings Cronbach's

alpha Factor 1 Factor 2

STI

Intramural R&D % of sales 0,9014 0,0488

0,7299 Intramural R&D % of total personnel 0,877 0,0941

External R&D % of sales 0,6195 0,0259

DUI

New business

practices 0- No / 1- Yes 0,0541 0,8335

0,7087 New methods of

organising work 0- No / 1- Yes 0,0354 0,8503

New methods of

organising external 0- No / 1- Yes 0,1529 0,684

Percentage of variance explained 0,3322 0,3163

Control variables Measurement

Sales size Ln(Salest-1)

Enterprise group 0- No / 1- Yes

Manufacturing firm 0- No / 1- Yes

Technological firm 0- No / 1- Yes

Control variables Measurement

Spanish firm 0- No / 1- Yes

Private firm 0- No / 1- Yes

Firm age Age in years

Science park 0- No / 1- Yes

Description of variables

1. Organizational performance: • Annual sales change:

2. Innovation output: • Product innovation

• Process innovation

Dependent

variables

(Yit)

100 Sales

Sales= Sales

1-ti,

ti,

ti,

Standardized

scale indicators

Construct Variables Measurement Factor loadings Cronbach's

alpha Factor 1 Factor 2

Product

New goods 0- No / 1- Yes 0.8401 0.1673

0.7407 New services 0- No / 1- Yes 0.5240 0.2537

New products 0- No / 1- Yes 0.9139 0.2104

New to market % of sales 0.6316 -0.0524

Process

New prod. methods 0- No / 1- Yes 0.2939 0.6767

0.7502 New logistic systems 0- No / 1- Yes 0.0723 0.6125

New support act. 0- No / 1- Yes 0.0786 0.7611

New processes 0- No / 1- Yes 0.2072 0.8834

Percentage of explained variance 0.2944 0.2915

Results: Impact of STI and DUI on innovationa

Product Process

(1) (2) (1) (2)

STI i,t-1 0.139*** 0.063***

DUI i,t-1 0.076*** 0.125***

STIxDUI i,t-1 -0.016** -0.012*

STI i,t-2 0.120*** 0.052***

DUI i,t-2 0.078*** 0.132***

STIxDUI i,t-2 -0.014* -0.012

Observations 33,789 25,711 33,789 25,711

Cases 8,681 8,407 8,681 8,407

R2:

Within 0.008 0.002 0.008 0.002

Between 0.312 0.284 0.312 0.284

Overall 0.263 0.259 0.263 0.259

a Coefficients for control variables omitted due to space constraints

(STI > DUI) Prob > X2: 0.000 0.000 0.000 0.000 (DUI > STI) Prob > X2:

Results: Impact of innovation on performancea

Sales growth

(1) (2)

Producti,t-1 0.500*

Processi,t-1 0.603**

Producti,t-2 0.831***

Processi,t-2 0.582**

Observations 33,789 25,711

Cases 8,681 8,407

R2:

Within 0.182 0.200

Between 0.025 0.042

Overall 0.135 0.140

a Coefficients for control variables omitted due to space constraints

0.742 0.498 (Poduct > Process) Prob > X2: (Process > Product) Prob > X2:

Conclusions

1. Conducting STI and DUI activities is a complex and risky task – The combined effect does not seem to improve

each of the separate effects of STI and DUI

2. STI and DUI exert a positive effect on innovation – Stronger impact of STI on product innovation

– Stronger impact of DUI on process innovation

3. Potential short-term impact of process innovation, potential long-term impact of product innovation, but not statistically significant.

Implications

• Benefits from both product and process innovation

• Innovation policy actions should be concentrated not only on STI indicators, but also on DUI ones: – Experience-based knowledge driven by interaction of

agents within and out of the firm is also important

– But STI and DUI simultaneously may be challenging

• Despite the general evidence supporting subsidies to innovation, specific additionality of innovation programs should be measured and carefully assessed.

Many thanks!!

Any comment?

[email protected]

[email protected]

[email protected]