Eurasian Business Review, 3(1), 2013, 47-65

INNOVATION AND PRODUCTIVITY: AN UPDATEW

Pierre Mohnen* and Bronwyn H. Hall**

Abstract: This paper reviews the existing evidence regarding the effects of
technological and non-technological innovations on the productivity of firms and the
existence of possible complementarities between these different forms of
innovation.

Keywords: Innovation, Productivity

JEL Classification: O30

1. Introduction

In the minds of many people, and certainly in the view of most policy
makers, innovation is a key factor of economic growth. Innovation can be
divided into technological innovations in the form of new products and
services and non-technological innovations in the form of organizational or
marketing changes. Growth itself can be achieved by putting more factors
of production to work (increased investment, use of more land, decrease in
unemployment and increase in labor force participation) and by achieving
higher levels of output with the same amount of resources (total factor
productivity -TFP- growth). Innovation per se does not increase the amount
of productive resources, hence it affects growth mainly through TFP. By
which channels does innovation affect TFP? What evidence do we have to
state that innovation increases TFP? What kind of innovation has the
greatest impact on TFP? Is there a complementarity between different
forms of innovation? Those questions will be the main object of this paper.
This survey of the literature updates the survey by Hall (2011) on
innovation and productivity and complements the Mairesse and Mohnen
(2010) survey on the use of innovation surveys to better understand
innovation.
The paper is organized as follows. In sections 2 and 3 we define
respectively the notions of innovation and productivity, and we discuss the
way they are measured. In section 4 we explain how, why and when
innovation is likely to affect productivity. In section 5, we describe how the
link between innovation and productivity has been modeled in empirical
studies. In sections 6 and 7 we discuss the evidence gathered so far
regarding the link between innovation and productivity and possible
complementarities between different forms of innovation. Section 8
concludes.

W
This paper was in part produced as part of the SCIFI-GLOW Collaborative Project
supported by the European Commission’s Seventh Framework Programme for Research
and Technological Development, under the Socio-economic Sciences and Humanities
theme (Contract no. SSH7-CT-2008-217436). We thank Marco Vivarelli and Ender Demir for
their valuable suggestions and comments.
* Corresponding Author: Maastricht University and UNU-MERIT, Maastricht, The
Netherlands. Email: [email protected]
** Maastricht University, UNU-MERIT, University of California at Berkeley, and NBER.
Email: [email protected]
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

2. What is Innovation and How is it Measured?

Innovation can be measured by its inputs (the efforts made by firms to

come up with new products, new ways to produce their output or to run
their business more efficiently and to conquer new markets) or by its output
(new products or processes successfully introduced, increases in profits or
efficiency). On the input side, the first measure that comes to mind is R&D.
But performing R&D is not enough to be successful in bringing a new
product on the market. Innovation expenditures also encompass the
acquisition of machinery, equipment and software to produce new products
or processes, the purchase or licensing of patents, training related to the
introduction of new products or processes, market research, and feasibility
studies.
OECD’s Oslo Manual (2005) sets the guidelines for the innovation
surveys that collect data on innovation outputs, inputs and modalities. On
the output side it distinguishes four types of innovation: product, process,
organizational and marketing innovation. More formally they are defined as
follows: “A product innovation is the introduction of a good or service that
is new or significantly improved with respect to its characteristics or
intended uses. This includes significant improvements in technical
specifications, components and materials, incorporated software, user
friendliness or other functional characteristics. A process innovation is the
implementation of a new or significantly improved production or delivery
method. This includes significant changes in techniques, equipment and/or
software. An organizational innovation is the implementation of a new
organizational method in the firm’s business practices, workplace
organization or external relations. A marketing innovation is the
implementation of a new marketing method involving significant changes in
product design or packaging, product placement, product promotion or
pricing.”1 As indicated in the Oslo Manual, the borderlines of these
definitions can be debatable. Products are to be understood as goods or
services. Design changes which do not affect the functionalities or intended
uses of the product do not qualify as new products but do qualify as
marketing innovations. A new product may require new production
technology. A new product can at the same time be a marketing innovation,
when the functionalities or uses of the product change but also its external
appearance. A new method of producing a good, i.e. a process innovation,
may automatically involve a reorganization of work within the enterprise.
Innovation in its different forms can most easily be measured by a
dummy variable. But this measure does not adequately measure the extent
or intensity of innovation. For product innovation, the extent of innovation
within the firm can be measured by the share of total sales that is due to
new products. For process innovation, a few countries have chosen to
measure the extent of cost reduction brought about by process innovation.
For product, and in some countries also for process, innovations a

1
See OECD (2005), annex B, pp.149-154.

48
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

distinction can be made between “new to the firm” or “new to the market”,
depending on whether it is new only to the firm but already existing in the
market or whether it corresponds to a product or process that did not exist
before on the market. “New” can also be articulated as entirely new,
substantially improved or marginally improved. It goes without saying that
these notions do make economic sense but are difficult to measure in
practice. Innovation surveys deliver data that are to a large extent
subjective.
There are other forms of innovation that we shall not consider in this
brief survey. First, we shall not look into the much used alternative output of
innovation, or rather inventive activity, patents, which are used as formal
means of protecting intellectual property rights associated with invention.
Second, as mentioned in the beginning, innovative effort can also be
measured on the input side, by R&D or other innovation expenditures. We
shall only look at the relationship of the innovation output measures to
productivity.2 Third, innovations can be classified according to the initiator
of the innovation: the public sector (public innovations), the user (user
innovations), and innovations introduced by communities, which are often
user innovations based on traditional knowledge, called “grassroot
innovations”. Other ways to categorize innovations are as innovations in the
way society is organized (social innovations), innovations for the poor, also
denominated as “inclusive innovations” or “pro-poor innovations”, and
finally innovations with an environmental objective (environmental
innovations). We shall only include those innovations if they appear in the
form of one of the four innovations we have mentioned in the previous
paragraph.

3. What is Productivity and How is it Measured?

Suppose you had only one input, labor (L), to produce a certain amount of
production (Q). Production would increase if more labor was hired and put
to work. But it could also increase if labor was used more efficiently or if a
new technology was adopted that raised the amount of output per labor, so-
called labor productivity. Likewise with multiple factors of production, more
could be produced by putting more units of each factor to work or by
increasing the amount produced with the same amount of inputs. Again it
could be due to a change in efficiency, which could partly be due to a
substitution between inputs, e.g. a higher capital/labor ratio, or the adoption
of a new technology. In a multi-input, multi-output context, productivity is
defined as the ratio of an index of output over an index of input.
A first difficulty in measuring productivity is how to construct these
indexes. The basic idea is that each factor gets a weight corresponding to
its individual contribution, so that a more productive factor gets a higher
weight than a less productive factor. If we knew the exact functional form of

2
For a recent survey of the relationship between R&D spending and productivity, see Hall et
al. (2010).

49
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

the production function, we could construct exact indexes (Diewert, 1976).

These individual contributions can either be estimated econometrically or, if
we can assume that markets are competitive, factors are always adjusted
to their optimal levels and returns to scale are constant, the individual factor
returns can be approximated by their factor prices normalized by the price
of output.

3.1. How Can Productivity Increase?

A first explanation lies in the exploitation of scale economies, the output

expanding more than the inputs. A second explanation is the presence of
unused capacity utilization. If some machines stay idle or there is
temporarily excess labor, then production can partially be increased without
hiring additional inputs. This situation refers to the cyclical nature of
productivity and the presence of rigidities in input markets (labor hoarding,
indivisibilities in capital stock, adjustment costs, time to build). A third
explanation is technological change, i.e. new ways of producing old things
requiring overall less input per unit of output. This outward shift of the
production frontier corresponds to a new best practice. But firms can also
get closer to the best practice by investing in new machines or by adopting
new management techniques, something that in the literature is termed
change in efficiency (Farrell, 1957).

3.2. The Measurement or the Estimation of Productivity

The measurement or the estimation of productivity is full of challenges.

Besides the issues of assigning a different marginal return to every input,
depending on its quality (e.g. distinguishing workers by their skill levels), or
the issue of separability between primary and intermediate inputs (i.e.
bringing the non-separability between value added and intermediate inputs
into the picture), there are two main measurement challenges that are
particularly related to innovation. One is the incorporation of quality
changes. If the output quality improves without fully showing up in the price
statistics, then nominal output gets deflated too much when using an
industry-wide output deflator. In this case, the quality improvement shows
up as increases in revenue (price times quantity) but not as increases in
real output. The same can be said on the input side. If for instance ICT
equipment, which underwent huge quality improvements in the last twenty
years, still gets deflated at the old prices that do not include quality
adjustments, then input is undervalued and hence TFP in the using sector
is overestimated. So the choice to quality adjust the price of ICT affects the
allocation of productivity gains between the producing and using sectors.
The other challenge has also to do with prices but not as they relate to
quality but as they relate to non-competitive pricing. Typically, innovators
have for some time a market-power position that allows them to sell their
products or services at above competitive prices, as in monopolistic
competition, and hire some of their inputs, e.g. high-skilled labor, at below

50
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

competitive prices, as in monopsonistic competition. Again, if output

deflators are undervalued and input deflators overvalued, then part of
productivity reflects price effects (see for instance Dobbelaere and
Mairesse, 2010).

4. How does Innovation Affect Productivity?

Putting a new product on the market creates a new source of demand,

which can give rise to scale economies in its production or to improved
productivity because its production requires less of the inputs than the old
products, although now we are already implicitly assuming a new
production process or technology for the production of the new product.
The new products may of course cannibalize the business and the profits
made from producing the old products when the new products replace and
drive out the old products from the market. The contrary may happen when
the new products are complementary to the existing products. It is possible
that selling the new products in parallel to the old products may lead to
economies of scale in the distribution of the goods on the market. It may
also be that at the beginning productivity declines, and afterwards it
improves as the firm moves down the learning curve. Among the new
products launched, some may be more successful than others, because
they satisfy an immediate or latent need for the customers, or they benefit
from a me-too snowball effect, or they nicely complement some other newly
introduced product or service on the market.
Process innovation is a priori expected to have a clearer positive
effect on productivity as new processes are often introduced in order to
reduce production costs by saving some of the more costly inputs (often
labor). Besides the direct effects on productivity, innovations can also have
indirect effects, as when an initial productivity improvement leads to a price
reduction, which, if demand is sufficiently price responsive, leads to a more
than proportional increase in sales, which can create additional productivity
improvements in the presence of returns to scale. The extent to which the
unit cost reductions get translated into a price decrease depends on the
extent of competition in the market, which can itself be a function of how
important the innovation is.
The importance of a given product innovation can also be measured
by the degree of novelty. A product new to the firm but not to the market
can be regarded as a minor innovation, some would even qualify it as an
imitation, whereas a product new to the market represents a more drastic
innovation. In some surveys, like the Canadian survey, separate
geographic markets are considered, like the provincial, national, North-
American and world markets. A new to the market product has a larger
potential for success. If it can be sold rapidly on a large market and if it
corresponds to customer needs in all parts of that large market, by its sheer
size it can benefit from scale effects and improve productivity. Competition,
however, is likely to be stronger on the world market than on a local market

51
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

and so is the danger of imitation. It is thus not immediately clear whether

the scale effect or market power and appropriability effects dominate.
The success of a product on the market may depend on the quality
of the associated marketing or on the (re)training of workers that produce
the product. The productivity effects of innovation may therefore depend on
the simultaneous presence of various types of innovation and it is
interesting to investigate the presence of complementarity between
different innovation modes.

5. Modeling the Link between Innovation and Productivity

Most of the models that have estimated the impact of innovation on

productivity have done so within the so-called Crépon et al. (1998) – CDM
– model. This model is generally presented as a recursive system of three
blocks of equations. A first block explains the determinants of the
probability to do R&D and of the intensity of R&D:

‫ݎ‬ଵ ൌ ͳሾ‫ݎ‬ଵ‫ כ‬൐ Ͳሿ where ‫ݎ‬ଵ‫ כ‬ൌ ࢄଵ ࢼଵ ൅ ߝଵ

‫ݎ‬ଶ ൌ ‫ݎ‬ଶ‫ כ‬ൌ ࢄଶ ࢼଶ ൅ ߝଶ if ‫ݎ‬ଵ‫ כ‬൐ Ͳ and zero otherwise

where ‫ݎ‬ଵ is the indicator variable indicating whether there is R&D or not, ‫ݎ‬ଶ
is the intensity of R&D, ࢄଵ and ࢄଶ are vectors of explanatory variables and
the ε’s are the error terms.3

A second block explains the determinants of the probability to be

innovative, in one way or the other, and the extent of product innovation
(and/or process innovation if the data permit), R&D being one of those
determinants:
݅ଵ ൌ ͳሾ݅ଵ‫ כ‬൐ Ͳሿ where ݅ଵ‫ כ‬ൌ ࢃଵ ࢽଵ ൅ ߟଵ
݅ଶ ൌ ݅ଶ‫ כ‬ൌ ࢃଶ ࣁଶ ൅ ߟଶ if ݅ଵ‫ כ‬൐ Ͳ and zero otherwise,

where ݅ଵ is the indicator variable indicating whether there is innovation

output or not (e.g. product innovation), ݅ଶ is the intensity of innovation
output, ࢃଵ and ࢃଶ are vectors of explanatory variables, one component of
which is ‫ݎ‬ଵ‫ כ‬or ‫ݎ‬ଶ‫ כ‬ǡor their observed equivalent, and the η’s are the error
terms.
Finally the productivity equation depends on innovation output
(݅ ൌ ݅ଵ‫ כ‬, ݅ଶ‫ כ‬, ݅ଵ or ݅ଶ ), besides other explanatory variables Z (like physical
capital intensity):
ܳ
ൌ ࢆμ ൅ ݅߶ ൅ ‫ݑ‬Ǥ
‫ܮ‬
Generally, there is no feedback from productivity to R&D or to
innovation, the model is static, it is estimated on cross-sectional data, the
productivity is estimated in levels, and R&D does not enter productivity

3
For simplicity we ignore the individual or time subscript.

52
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

directly.4 When R&D and innovation are only measured as dichotomous

variables, it is the incidence and not the intensity of innovation then enters
the productivity equation.
The interesting feature of the CDM model is that it handles some of
the endogeneity of R&D and innovation, in the innovation and productivity
equations respectively, and that it explicitly models the selectivity of R&D
performers and/or innovators. As far as the estimation is concerned, there
is the usual tradeoff between efficiency and robustness to misspecification.
The original model was estimated by asymptotic least squares, or minimum
distance estimator, where all equations are estimated jointly. Even more
information is exploited when the model is estimated by maximum
likelihood with given distributions for the random part of the model. Most
studies have opted for a sequential approach, where the predicted value of
one endogenous variable enters the estimation of the next equation, with
due account for the calculation of the standard errors and the inclusion of a
correction factor for potential selection bias. Hall et al. (2009) and Musolesi
and Huiban (2010) do not report a great difference in the estimation results
when comparing a sequential IV estimation with a maximum likelihood
estimation approach. As long as the endogeneity and selection are
somehow treated, the results are rather robust to the estimation method.
But, as illustrated in Mairesse et al. (2005), when endogeneity or selectivity
are not corrected for, the significance of the estimated parameters drops
tremendously, pointing to an error in variables problem, probably related to
the subjectivity of the answers to some of the questions that generated the
data, rather than a simultaneity problem.

6. Technological versus Non-Technological Innovation

Our first interest is to investigate the size of the elasticity of productivity

(labor or total factor) with respect to the intensity of innovation. Most
innovation surveys measure the intensity or the success of product
innovation by means of the share of new products in total sales, i.e. the
proportion of total sales that is due to products launched in the last three
years (according to the Oslo Manual).5 Some innovation surveys, such as
the Swiss innovation survey, also try to capture the intensity of process
innovation by asking the percentage of cost reductions due to process
innovations made in the last three years. The other types of innovation are
only captured by dummy variables, given the difficulties of measuring their
specific contribution to output.
Table 1 summarizes a number of empirical studies that have
estimated the elasticity of productivity with respect to the intensity of
product innovation. The elasticities are, but for one case, positive and in

4
These limitations reflect the limitations of the usual innovation surveys, which draw a new
sample for each edition, precluding any panel data analysis.
5
In the annual industrial survey organized by the China National Bureau of Statistics new
product sales cover the products introduced in the year covered by the survey.

53
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

most cases significant. The magnitude of the elasticity varies, but it is not
uncommon to find elasticities of the order of 0.25, implying that if innovative
sales per employee go up by 10%, labor productivity rises by 2.5%. The
magnitudes are lower and more volatile when the elasticity concerns the
share of total sales due to new products instead of sales of new products
per employee. They also tend to be lower when the growth rather than the
level of productivity is estimated and when skilled labor or human capital is
controlled for (Crépon et al. 1998; Criscuolo, 2009). In the only case where
the elasticity had a negative sign (in Roper et al. 2008) knowledge capital
utilization in the form of skilled labor was controlled for. Therrien and Hanel
(2009) in their report of a few extensions of the core OECD model also
remark that the introduction of human capital, physical capital and the use
of value added per employee rather than sales per employee tend to
reduce the productivity elasticity of output. This result suggests an
identification problem between innovation and other measures of
knowledge capital and physical capital. In the countries where services
sector data were available, the OECD study led by Criscuolo (2009) reports
that the effect was generally higher for manufacturing than for services
firms with the notable exceptions of Germany and New Zealand. Lööf and
Heshmati (2006) and Mairesse et al. (2005) do not find a significant
difference in the elasticity of productivity with respect to the intensity of
product innovation when they distinguish between products new only to the
firm and products new to the market.6
Unfortunately, for all other types of innovation – process,
organizational and marketing – the only innovation measures available are
dichotomous measures. These measures are less satisfactory because first
they refer to a three-year period (whereas the intensity refers to the last
year of this three-year period) - so it is not clear what the exact timing is -,
second they refer to various projects without weighting them by their level
of success – blockbusters are mixed with flops - , and third they do not
correct for size – it is normal than larger firms with more projects will have a
higher chance to be innovative with at least one of them. But nonetheless
they should give us some indication of the differential effect of various types
of innovation on productivity. We shall in particular distinguish technological
(product and process) from non-technological (organization and marketing)
innovations.

6
For Lööf and Heshmati (2006), see their table X. These results are not reported in our
Table 1.

54
 Table 1. Studies on innovation and productivity using a continuous measure of product innovation
Authors Country Observations Estimation Output measure Innovation measure Impact of innovation+ Additional comments
(year) method
Crépon et al. France 4164 manufacturing firms, ALS Log value added per Log share of innovative 0.104*** (0.016) Control for capital

P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65
(1998) 1986-1990 employee sales stock/employee;
0.065*** (0.015) + control for labor skill
Lööf et al. Finland 353 mfg firms, 3SLS Log sales per Log innovative sales 0.090 (0.058) Control for process
(2003) 1994-1996 employee per employee innovation dummy
Norway 485 mfg firms, 0.257*** (0.062)
1995-1997
Sweden 407 mfg firms, 0.148*** (0.044)
1994-1996
Janz et al. Germany 352 firms, Sequential IV + Log sales per Log sales income from 0.268*** (0.100) Control for process
(2003) 1998-2000 IMR employee product innovation per innovation dummy
Sweden 206 firms, employee 0.290*** (0.084)
1998-2000
(in knowledge-intensive
manufacturing)
Mairesse et France 889 firms in HT sectors, ALS Log sales per Logit transformation of 2.03 Control for capital stock
al. (2005) 1998-2000 employee share of innovative and materials per
1354 firms in LT sectors, sales 0.52** employee
1998-2000
55

Benavente Chile 438 manufacturing ALS Log Value added Log share of innovative 0.179* (0.113) Control for capital
(2006) plants,1995-1998 per employee sales per employee stock/employee
Lööf and Sweden 1974 manufacturing firms, 3SLS + IMR Log value added per Log innovation sales 0.121*** (0.043) manuf. Control for process and
Heshmati 1996-1998 employee per employee 0.093** (0.047) Services organizational
(2006) innovation
1081 service firms, 1996- Growth rate innov. 0.070 *** manuf.
1998 sales per employee 0.080** services
van Leeuwen Netherlands 1926 firms, 3SLS Growth of sales per Log innovative sales 0.133*** (0.026) Innovation not significant
and Klomp 1994-1996 employee per employee in growth of VA per
(2006) employee
Jefferson et China 5451 large and medium Sequential IV Log gross output Log share of innovative 0.035*** (0.002) Control for capital stock
al. (2006) sized mfg firms, sales and materials
1995-1999
Roper et al. Ireland and Panel of 1620 Sequential IV Value added per Share of innovative -0.302*** (0.067) Control for process
(2008) Northern observations over 4 employee sales innovation dummy, labor
Ireland innovation survey waves, skill
1991-2002
Criscuolo 17 OECD Micro data, 2002-2004, Sequential IV Log sales per Log innovative sales Between 0.3 and 0.7 Control for process
(2009) countries except for Austria (1998- employee per employee (mostly ***) innovation dummy
2000), Australia (2003-
2005), New Zealand
(2004-2005)
 Table 1. Continued
Siedschlag et Ireland Panel of 723 firms over Sequential IV Log sales per Log innovative sales 0.093*** mfg
al. (2010) two CIS waves, 2004- employee per employee 0.098*** services
2008
Mairesse et China 13245 firms in 4 Sequential IV Log sales per Log innovative sales Between 0.246*** and Estimated separately for

P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65
al. (2012) industries in 2005 and employee per employee 1.119*** each industry
2006
Raymond et France and Panel data, three waves Maximum Log sales per Logit transformation of From 0.043*** to 0.107*** Similar results when
al. (2012) of innovation surveys: likelihood employee share of innovative in France using observed or latent
Netherlands 1994-96, 1998-2000, sales From 0.045*** to 0.197*** occurrence or intensity
2002-04; 2505 in the Netherlands
observations in France
and 1639 in Netherlands
When more than one figure was reported, we took the preferred estimates, as reported by the authors. ALS stands for asymptotic least squares, IV for instrumental
variables, IMR for inverse Mill’s ratio; + standard errors in parentheses; mfg stands for manufacturing; ***significant at 1%; **significant at 5%; *significant at 10%
56
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

Table 2 summarizes those pieces of work that have estimated

semi-elasticities of productivity with respect to innovation dummy
variables.7 The first thing to note is that, when estimated separately (i.e.
without the other innovation dummy and without the intensity of innovation),
process or product innovations are significant (see Mairesse et al. 2005;
Parisi et al. 2006; Raffo et al. 2008, and Siedschlag et al. 2010). When
innovation intensity is controlled for, it often happens that the coefficient of
process innovation is negative and significantly so (Janz et al. 2003; Lööf
and Heshmati, 2006; van Leeuwen and Klomp, 2006; Criscuolo, 2009).
When product and process innovation dummies appear together, their
coefficients often turn out non-significant, and if one of the two is significant
it is more often product innovation (Griffith et al. 2006; Mairesse and Robin,
2008; Musolesi and Huiban, 2010). There seems to be again an
identification problem there. The stronger measure of innovation (the
intensity of product innovation) dominates the more noisy process
innovation dummy. It could also be argued, see Hall (2011) that product
innovations create a market power effect that increases the revenue
measure of output, whereas efficiency improvements from process
innovations may not show up in the revenue figures if they result in lower
prices without corresponding increases in output (at least in the short run).
Another identification problem could be due to the fact product and process
innovations often appear together and that only their joint effect is the most
visible (see Hall et al. 2009 and to some extent Chudnovsky et al. 2006).
Masso and Vahter (2008) have compared the productivity effect of
various kinds of innovation occurring during a three-year period on the
productivity observed at the end of that period and one or two periods
ahead. Results are not significantly different. But, Huergo and Jamandreu
(2004), in their preferred specification of TFP growth on age and process
innovation on a panel of Spanish firms between 1990 and 1998, find that
the process innovation dummy increases TFP growth by 1.5% the year of
implementation followed by a three year long lower TFP increase, and then
a strong TFP decline if no new process innovation takes place. Raymond et
al. (2012) allow for persistence in innovation and productivity and for a
feedback from productivity on product innovation (occurrence or success).
They find signs of a Granger causality from past innovation on current
productivity but not from past productivity on current innovation.

7
Some early studies on innovation and productivity have used the number of innovations
from the SPRU database. Sterlacchini (1989) obtained on a panel of 15 Italian
manufacturing industry data a coefficient of 0.08 (0.04) for the number of innovations
produced in a long-run TFP growth regression but no significant coefficient for the number of
innovations used. Geroski (1989) reports a coefficient of 0.025 (0.010) for the number of
innovations introduced in the last three years on a panel of 79 UK industries.

57
 Table 2. Studies on innovation and productivity using dummy variables for various kinds of innovation
Authors (year) Country Observations Estimation Output Innovation Impact of Additional comments
method measure dummies innovation
Janz et al. (2003) Germany 352 firms, Sequential Log sales per Process -0.136 ** (0.069) Innovation intensity is controlled

P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65
1998-2000 IV + IMR employee for
Sweden 206 firms, -0.030 (0.119)
1998-2000
(from knowledge-intensive
manufacturing)
Huergo and Spain Panel 2300 firms 1990-98 Non- Solow Process 0.015 *** Positive immediate effect that
Jaumandreu (2004) parametric residual TFP declines afterwards and become
growth <0 after 3 years without new
innovation
Mairesse et al. France 889 firms in HT sectors, ALS Log sales per Product new to firm 0.031*** HT Introduced separately
(2005) 1998-2000 employee Product new to 0.051 *** LT
market 0.047 *** HT
1354 firms in LT sectors, Process 0.050 *** LT
1998-2000 0.063 *** HT
0.097 ** LT
Lööf and Heshmati Sweden 1974 manufacturing firms, 3SLS + IMR Log value Process -0.071 *** mfg Innovation intensity is controlled
(2006) 1996-98 added per -0.071 services for
58

1081 services firms, 1996-98 employee Organizational -0.027 mfg

-0.069 services
Parisi et al. (2006) Italy 465 manufacturing firms, Sequential labor Product 0.08 (0.054) 3 period growth rates; when
1992-1997 IV productivity Process 0.14 (0.054)*** both are introduced, they often
growth turn non-significant
459 manufacturing firms, Sequential TFP growth Product 0.13 (0.069)*
1992-1997 IV Process 0.15 (0.047)***
Duguet (2006) France 4085 mfg firms, 1986-1990 ALS and TFP growth Radical and 0.022 (0.004)**
GMM incremental product for radical
innovation -0.01 (0.01)
for incremental
van Leeuwen and Netherlands 1926 firms, 3SLS Growth of Process -1.256 (0.471)*** Control for capital/employee and
Klomp (2006) 1994-1996 sales per for share of innovative sales
employee
Griffith et al. (2006) France 3625 firms, Sequential Log sales per Product and process 0.060*** prod
1998-2000 IV employee 0.069** proc
Germany 1123 firms, -0.053 prod
1998-2000 0.022 proc
Spain 3588 firms, 0.176*** prod
1998-2000 -0.038 proc
UK 1904 firms, 0.055*** prod
1998-2000 0.029 proc
 Table 2. Continued
Chudnovsky et al. Argentina Panel of 718 mfg firms over 2 Sequential Log sales per Product only 0.088 (0.076)
(2006) innovation survey waves, estimation employee Process only 0.178 (0.081)**
1992-1996 , 1998-2000 with fixed Prod + proc 0.136 (0.055)**
effects

P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65
Masso and Vahter Estonia 1142 firms, 1998-2000 Sequential Log of value Product, process, 0.002 prod Control for capital intensity;
(2008) 916 firms, 2002-2004 IV added per organization 0.151*** proc bivariate probit
employee 0.097* org
Raffo et al. (2008) France (FR) Cross-sectional data, Sequential Log sales per Product and 0.075***(FR) prod Organizational dummy
Spain (ES) 1998-2001, except for Spain IV employee organization 0.156*** (ES) prod significant only for Brazil
Switzerland (CH) (2002-2004) 0.101* (CH) prod
Argentina (AR) -0.219 (AR)prod
Brazil (BR) 0.220*** (BR)prod
0.054***(BR)org
Mexico (MX) 0.313***(MX)prod
Roper et al. (2008) Ireland and Panel of 1620 observations Sequential Value added Product 0.011 (0.031) Control for innovation success,
Northern Ireland over 4 innovation survey IV per employee Process 0.008 (0.030) labor skill, capital intensity
waves, 1991-2002
Mairesse and France 3524 manuf firms, 1998- Maximum Log value Product and process 0.14 *** product Bivariate probit
Robin (2008) 2000 likelihood added per 0.02 process
4955 manuf firms, 2002-2004 employee 0.13 *** product
3599 services firms, 2002- -0.02** process
59

2004 0.17*** product

-0.01 process
Hall et al. (2009) Italy 9674 firm-year observations, Sequential Log sales per Product and process 0.193 (0.267) Bivariate probit; process
panel of 10 years, mostly IV employee process only innovation becomes non-
SMEs, 1992-2003 significant if
0.597*** (0.093) investment/employee is
product included (reported figures);
without investment included,
process innovation dominates
Musolesi and France 416 knowledge intensive Maximum Log value Product 0.324***(0.124) Has also been estimated by
Huiban (2010) business service firms, 1998- likelihood added per Process 0.131 (0.198) 2SLS, and IV
2000 employee Technolog. 0.210* (0.117)
Non-technolog. 0.271 (0.194)
Criscuolo (2009) 17 OECD countries Micro data, 2002-2004, Sequential Log sales per process <0 or non-significant Control for continuous product
except for Austria (1998- IV employee innovation
2000), Australia (2003-2005),
New Zealand (2004-2005)
Siedschlag et al. Ireland Panel of 723 firms over two Sequential Log sales per Mfg services Not estimated jointly
(2010) CIS waves, IV employee . Product 0.257** 0.609***
2004-2008 . Process 0.213** 0.450***
.Organization 0.373***1.058***
***significant at 1%; **significant at 5%; *significant at 10%; not all estimation results are always reported.
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

Masso and Vahter (2008) report that they do not find a significant
difference in the innovation semi-elasticities when measuring productivity in
terms of sales per employee or value added per employee, and that the
effect of the various forms of innovation tend to be non-significant if the
dependent variable is productivity growth. Duguet (2006) reports that only
new-to-the-market product innovations have a significant effect on TFP
growth.
Greenan and Guellec (1998) have shown that what we would now
call organizational and marketing innovations had a positive effect on total
factor productivity in a cross-section of French firms in 1987. Black and
Lynch (2004) show that workplace innovations like reengineering,
incentivizing, profit-sharing, have raised total factor productivity in US
manufacturing establishments between 1993 and 1996. A few recent
studies (Masso and Vahter, 2008; Polder et al. 2009; Musolesi and Huiban,
2010) have introduced the organizational or non-technological innovation
dummies in productivity regressions. The results are similar to those
obtained for product and process innovations, and the same critical
remarks apply.

7. Complementarity between Different Types of Innovation

A new product may require a new way of producing it with lighter materials
but a need for more precision instruments in the fabrication of the new
product. Product innovations may thus often be combined with process
innovations. New production processes in turn may raise productivity only if
they are combined with a reorganization of work. On the one hand, ICT
allows more decentralized decision making but also requires a higher
degree of integration of the different activities, for instance through the use
of an enterprise resource planning software (see Bresnahan et al. 2002;
and Crespi et al. 2006). The introduction of a new way of producing a given
product or service may thus need to be accompanied by an organizational
innovation. The success of a new product or process on the market may
depend on the quality of advertising, the speed in bringing it to the market,
efficiency in its distribution, and after-sales service. In other words, product
innovations may be more successful if complemented by marketing
innovations.
Complementarity between two or more variables (often called
strategies) can be tested by checking whether the demand for one
increases in the presence of the other one (at least in the case of two
variables)8 or whether the joint use of two or more variables leads to a
higher performance. In the latter case, a performance measure needs to be
chosen. In the former case, the source of the complementarity remains
unexplained. It is important, whenever possible, to correct for time-invariant

8
In the case of more than two variables, the interdependence between all the variables
needs to be taken into account.

60
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

individual effects so as not to attribute the complementarity to individual

time invariant characteristics.
Martinez-Ros and Labeaga (2009) find evidence of complementarity
between product and process innovations in Spanish manufacturing
correcting for unobserved individual time-invariant heterogeneity. Product
(process) innovation in one year increases if process (product) innovation
occurred in the previous year. Miravete and Pernías (2006) found evidence
of complementarity between product and process innovation in the Spanish
ceramic tile industry.
The study of complementarity has been extended to organizational
innovations. Ballot et al. (2011) for the UK and France, and Polder et al.
(2009) for the Netherlands test the existence of pairwise and full
complementarities between product, process and organizational
innovations. Full complementarity is never obtained and the
complementarity between pairs of strategies depends on the country
examined, and, as shown by Ballot et al. (2011), it is contingent on the size
of the firms and their knowledge intensity. Product and process innovations
are found complementary in all three countries, product and organizational
innovations in France, and process and organizational innovations in the
UK and the Netherlands. In the Netherlands, the synergies were not found
to be different for manufacturing and for services firms. Both studies though
do not correct for unobserved individual effects.
Schmidt and Rammer (2007) obtain evidence on German firm data
that the success with the introduction of market novelties (not of products
new to the firm) or with process innovations increases in the presence of
organizational and marketing innovations. Polder et al. (2009) report that in
both manufacturing and services the combinations of innovations that
contribute significantly to a higher productivity all involve organizational
innovation: organizational innovation only, process combined with
organizational innovation, and the combination of all types of innovation.

8. Conclusion

We can conclude from this brief survey of the empirical literature on

innovation and productivity that innovation leads to a better productivity
performance, or to be more precise to a better revenue per employee
performance. Some of the effect of innovation goes to real output, and
some of it to the price at which the output is sold. In the absence of good
individual price measures it is hard to dissociate these two effects.
All four types of innovations considered - product, process,
organizational and marketing innovations - contribute to a better
productivity performance. Given the imperfect measurement of innovation
and the simultaneity of different types of innovation, it is difficult to isolate
the individual effect of each. Some complementarity between them seems
to exist, even though it is hard to get a good grasp of the exact nexus of
complementarities.

61
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

To progress in our understanding of the link between innovation and

productivity, a few avenues are worth exploring. First, quantitative data are
more likely to produce meaningful and robust results than qualitative yes/no
data. An effort could be made to construct quantitative data for other than
product innovations. Second, as more data become available it would
certainly be worth constructing a panel dataset that would allow to correct
for unobserved heterogeneity and to examine the dynamic aspects of the
relationship. Third, as much as the CDM model was an improvement over
the extended Cobb-Douglas production function approach for evaluating
the returns to R&D, it would enrich our analysis if we could set up a richer
structural model that would include the indirect aspects of innovation on
firm performance via price effects and competition. Fourth, it would be
interesting to analyze the entry and exit decisions of firms related to
innovation as well as the effect of uncertainty. Because of risk, innovation
may not just fail to show up in productivity figures but even lead firms to go
bankrupt. Finally, in a more macro-economic perspective the market exit,
competition and externality effects may yield quite a different picture of the
outcome of innovation than the micro-economic partial equilibrium analysis
pursued so far.

References

Ballot, G., Fakhfakh, F., Galia, F., and Salter, A., 2011. The fateful triangle
complementarities between product, process and organizational
innovation in the UK and France. TEPP Working Paper, 2011-5.
Benavente, J.M., 2006. The role of research and innovation in promoting
productivity in Chile. Economics of Innovation and New Technology,
15(4-5), pp.301-315.
Black, S.E. and Lynch, L.M., 2004. What's driving the new economy?: The
benefits of workplace innovation. The Economic Journal, 114(493),
pp.F97-F116.
Bresnahan, T.F., Brynjolfsson, E., and Hitt, L.M., 2002. Information
technology, workplace organization, and the demand for skilled
labor: Firm level evidence. Quarterly Journal of Economics, 117(1),
pp.339-376.
Chudnovsky, D., Lopez, A., and Pupato, G., 2006. Innovation and
productivity in developing countries: A study of Argentine
manufacturing firms behavior (1992-2001). Research Policy, 35(2),
pp.266-288.
Crépon, B., Duguet, E., and Mairesse, J., 1998. Research, innovation and
productivity: An econometric analysis at the firm level. Economics of
Innovation and New Technology, 7(2), pp.115-158.
Crespi, G., Criscuolo, C., and Haskel, J., 2006. Information technology,
organizational change and productivity growth: Evidence from UK
firms. CEPR Discussion Paper, 783.

62
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

Criscuolo, C., 2009. Innovation and productivity: Estimating the core model
across 18 countries. In: OECD, innovation in firms: A
microeconomic perspective. Paris: OECD Publishing. Ch.3.
Diewert, W.E., 1976. Exact and superlative index numbers. Journal of
Econometrics, 4(2), pp.115-145.
Dobbelaere, S. and Mairesse, J., 2010. Micro-evidence on rent sharing
from different perspectives. NBER Working Paper, 16220.
Duguet, E., 2006. Innovation height, spillovers and TFP growth at the firm
level: Evidence from French manufacturing. Economics of
Innovation and New Technology, 15(4-5), pp.415-442.
Farrell, M.J., 1957. The measurement of productive efficiency. Journal of
the Royal Statistical Society. Series A (General), 120(3), pp.253-
290.
Geroski, P.A., 1989. Entry, innovation and productivity growth. Review of
Economics and Statistics, 71(4), pp.572-578.
Greenan, N. and Guellec, D., 1998. Firm organization, technology and
performance: An empirical study. Economics of Innovation and New
Technology, 6(4), pp.313-347.
Griffith, R., Huergo, E., Mairesse, J., and Peters, B., 2006. Innovation and
productivity across four European countries. Oxford Review of
Economic Policy, 22(4), pp.483-498.
Hall, B.H., 2011. Innovation and productivity. Nordic Economic Policy
Review, 2, pp.167-204.
Hall, B.H., Lotti, F., and Mairesse, J., 2009. Innovation and productivity in
SMEs: Empirical evidence for Italy. Small Business Economics,
33(1), pp.13-33.
Hall, B.H., Mairesse, J., and Mohnen, P., 2010. Measuring the returns to
R&D. In: B.H. Hall and N. Rosenberg, eds. Handbook of the
economics of innovation. Amsterdam: Elsevier, pp.1034-1076.
Huergo, E. and Jaumandreu, J., 2004. Firms’ age, process innovation and
productivity growth. International Journal of Industrial Organization,
22(4), pp.541-560.
Janz, N., Lööf, H., and Peters, B., 2004. Firm level innovation and
productivity – Is there a common story across countries. Problems
and Perspectives in Management, 2, pp.184-204.
Jefferson, G., Bai, H., Guan, X., and Yu, X., 2006. R&D performance in
Chinese industry. Economics of Innovation and New Technologies,
15(4/5), pp.345-366.
Lööf, H., Heshmati, A., Apslund, R., and Naas, S-O., 2003. Innovation and
performance in manufacturing industries: A comparison of the
Nordic countries. International Journal of Management Research, 2,
pp.5-36.
Lööf, H. and Heshmati, A., 2006. On the relationship between innovation
and performance: A sensitivity analysis. Economics of Innovation
and New Technology, 15(4/5), pp.317-344.
Mairesse, J. and Mohnen, P., 2010. Using innovation surveys for
econometric analysis. In: B.H. Hall and N. Rosenberg, eds.

63
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

Handbook of the economics of innovation. Amsterdam: Elsevier,

pp.1130-1155.
Mairesse, J., Mohnen, P., and Kremp, E., 2005. The importance of R&D
and innovation for productivity: A reexamination in light of the 2000
French innovation survey. Annales d’Economie et de Statistique,
79/80, pp.489-529.
Mairesse, J., Mohnen, P., Zhao, Y., and Zhen, F., 2012. Globalisation,
innovation and productivity in manufacturing firms: A study of four
sectors of China. ERIA Discussion Paper, 2012-10.
Mairesse, J. and Robin, S., 2008. Innovation and productivity in France: A
firm-level analysis for manufacturing and services (1998-2000 and
2002-2004). In: SYMPOSIUM on innovation at firm level. Faculty of
Economics and Management, University of Barcelona, Spain 1 July
2008.
Martinez-Ros, E. and Labeaga, J.M., 2009. Product and process
innovation: Persistence and complementarities. European
Management Review, 6(1), pp.64-75.
Masso, J. and Vahter, P., 2008. Technological innovation and productivity
in late-transition Estonia: Econometric evidence from innovation
surveys. European Journal of Development Research, 20(2),
pp.240-261.
Miravete, E. and Pernías, J., 2006. Innovation complementarity and scale
of production. Journal of Industrial Economics, 54(1), pp.1-29.
Musolesi, A. and Huiban, J.P., 2010. Innovation and productivity in
knowledge intensive business services. Journal of Productivity
Analysis, 34(1), pp.63-81.
Organization for Economic Co-operation and Development (OECD), 2005.
Oslo Manual, 3rd ed. Paris: OECD.
Parisi, M.L., Schiantarelli, F., and Sembenelli, A., 2006. Productivity,
innovation creation and absorption, and R&D: Micro evidence for
Italy. European Economic Review, 50(8), pp.2037-2061.
Polder, M., van Leeuwen, G., Mohnen, P., and Raymond, W., 2009.
Productivity effects of innovation modes. Statistics Netherlands
Discussion Paper, 09033.
Raffo, J.D., L’huillery, S., and Miotti, L., 2008. Northern and Southern
innovativity: A comparison across European and Latin American
countries. European Journal of Development Research, 20(2),
pp.219-239.
Raymond, W., Mairesse, J., Mohnen, P., and Palm, F., 2012. Dynamic
models of R&D, innovation and productivity: Panel data evidence for
Dutch and French manufacturing. In: EEA & ESEM, 27th Annual
Congress of the European Economic Association & 66th European
Meeting of the Econometric Society. Malaga, Spain 27-31 August
2012.
Roper, S., Du, J., and Love, J.H., 2008. Modelling the innovation value
chain. Research Policy, 37(6-7), pp.961-977.

64
 P. Mohnen and B.H. Hall / Eurasian Business Review, 3(1), 2013, 47-65

Schmidt, T. and Rammer, C., 2007. Non-technological and technological

innovation: Strange bedfellows? ZEW Discussion Paper, 07-052.
Siedschlag, I., Zhang, X., and Cahill, B., 2010. The effects of the
internationalization of firms on innovation and productivity. ESRI
Working Paper, 363.
Sterlacchini, A., 1989. R&D, innovations, and total factor productivity
growth in British manufacturing. Applied Economics, 21(11),
pp.1549-1562.
Therrien, P. and Hanel, P., 2009. Innovation and productivity: Extending the
core model. In: OECD, Innovation in firms: A Microeconomic
perspective. Paris: OECD Publishing. Ch.4.
van Leeuwen, G. and Klomp, L., 2006. On the contribution of innovation to
multi-factor productivity growth. Economics of Innovation and New
Technologies, 15(4-5), pp.367-390.

65