The pursuit of variety:

the creation of new products and strategic differentiation

Nicoletta Corrocher * and Marco Guerzoni°

*KITeS, Bocconi University, Via Sarfatti 25, 20136 Milan, Italy

°Department of Economics and Business Administration, Friedrich Schiller Universität, Jena, Germany

Abstract

The paper examines the role of variety in a consumer goods sector – the ski

manufacturing industry - and its relevance in firms’ strategies. It empirically tests a

series of hypotheses on the relationship between the introduction of truly new

products and the development of small innovations around more standardized

products. We claim that the former is the true engine of growth and it may not be

compatible with the latter, which identifies firm-specific strategies of inventing

around existing products for the purpose of exploiting price differentials. Our results

show that, controlling for quality, highly innovative products are rarely introduced

by multi-product firms that have a differentiated product portfolio in terms of target

markets and technical characteristics.

*Corresponding author. Email: nicoletta.corrocher@unibocconi.it

1

I. Introduction

The generation of new varieties is crucial for economic growth. The capitalistic mode

of production has shown over time a remarkable capability of setting incentives for

economic actors to create new products and processes. Concerning the development

of new products, the incentives lie in the possibility of gaining market power either

by offering goods and services that are not provided by competitors, or by increasing

the range of product palette through small modifications and by exploiting the

resulting market segmentation in a strategic way. In terms of the impact on economic

growth, these two strategies are not equal, because the former generates something

truly new and therefore generates welfare, while the latter results in welfare transfers

among firms or between consumers and producers. Similarly, these two strategies

have a different impact on a firm’s organization of production. In particular, the

former requires a firm to give up economies of scale to a large extent and to make

targeted innovative efforts. In this paper we claim that firms cannot pursue both

strategies at the same time. Thus, we investigate whether the generation of truly new

products, which represents the real engine of growth, is hindered by strategies that

simply aim at introducing minor variations in existing products to exploit pricing

advantages.

Our empirical analysis relies upon an original product-level dataset including all

new skis introduced by 42 manufacturers and sold in the European market between

1992 and 2007. We collected data on the key product characteristics of 4202 models

and investigated the relationship between the novelty of the product and a series of

variables accounting for firm-specific differentiation strategies over time. In the next

paragraph, we clarify and discuss the two concepts of variety mentioned above and

we formulate three hypotheses. In the third section, we describe the dataset and

empirically test the hypotheses. Results and conclusions follow.

2

II. On the notion of variety

An economic system might take significant advantages from the existence of variety.

First, Saviotti (1991, 1994 and 2001) warns against the risk of a low production of

variety. He explains that if a system produces the same amount of output with a

decreasing amount of input (notably labour) due to productive gains from

standardization, it may not be sustainable in the long run. Indeed, the creation of

variety through both new products and new machineries is considered a necessary

condition to overcome technological unemployment. Second, David (1994)

admonishes against the possibility of a lock-in in the presence of standardization: the

existence of a standardized product can lead to important network effects and,

therefore, can generate high entry barriers in an industry, which in turn hinder the

diffusion of new products. Variety might be desirable also for many other reasons:

consumers can have a taste for variety because, as David (1994) emphasizes,

“…consumers may have a demand for intrinsic novelty as a means for combating the malaise

of boredom…”, or because they seek distinction (Swann, 2001), or because variety can

better fit their preferences (see Lancaster, 1990 for a review).

Notwithstanding the benefits of variety, its generation does not come with zero costs.

In other words, a real trade-off exists between standardization and variety: “…the

consumer gets lower costs but at the expense of variety” (Rae, 1985, p. 53). This trade-off is

a crucial source of industrial dynamics (Weitzman, 1992; David, 1994; David and

Rothwell, 1996). On the one hand, many benefits derive from standardization such as

a higher level of predictability of a product’s commercial success, faster learning

economies due to the simplification and routinization of the production process,

scale economies, an easier production of complementary assets and component

interfaces, and network externalities. Standardization allows an efficient

systematization of the production processes, by creating order and, consequently, by

reducing uncertainty. Moreover, product standardization does not necessarily

impinge on product quality, as is often believed. In contrast, standardization can lead

to more accuracy in the definition and production of standards. In the words of Rae

(1985, p. 53): “…the alleged sacrifice of quality to quantity is a myth”.

3

In discussing the issue of variety, the economic literature uses the notion of variety in

different ways (Lancaster, 1990): it can be the degree of novelty, i.e. its diversity from

the market average, or it can be the degree of product differentiation, i.e. the breadth

of a firm’s product palette. From an individual firm’s perspective, there are

incentives for pursuing both novelty – by introducing truly new products – and

differentiation – by increasing the number of products. Indeed, developing truly new

products can generate relevant competitive advantages for firms, because the more a

product differs from its competitors, the higher are the gains in market power (see

Hotelling, 1929). However, firms also have incentives to pursue differentiation and

increase the breadth of their product palette. As proved by existing theoretical

models, a multi-product firm might leverage on asymmetric pricing schemes and

have part of its product line shielded from competitors (Giraud-Héraud et al., 2003).

If each firm produces only one variety, such as in a pure Dixit and Stiglitz framework

(Dixit and Stiglitz, 1977), the production of each new variety requires the entry of a

new firm, which produces negative externalities on the other firms´ profits, by

increasing the competition. If on the contrary, firms are allowed to produce more

than one variety (up to an extreme case, where only one monopolist produces all the

possible varieties), they are able to internalize those negative externalities and thus

reach a more profitable equilibrium (see Vassilakis, 1992, among others, for a clear

theoretical discussion).

Although from an individual firm’s perspective there are incentives to pursue variety

by way of any of these two strategies, when we turn to examine the impact on

economic growth, the two concepts of variety generation – i.e. the degree of novelty

and the breadth of product differentiation – are not equivalent, because the latter

simply results in welfare transfers among firms, or between consumers and

producers, whereas the former generates something which is truly new. One might

therefore wonder whether there is a trade-off between these two strategies at the firm

level and, if so, what are the incentives for developing true innovations vs.

marginally differentiating the products.

4

Starting from these considerations, the aim of this paper is to investigate whether the

generation of truly new products, which is the real engine of growth, can coexist

with firm strategies that simply introduce minor variations with the aim of exploiting

pricing mechanisms.

The expected negative answer is found analysing the effect of the above mentioned

trade-off between product novelty and standardization at the firm level. The

manufacture of a product for a standard market allows firms to quickly exploit

learning economies, to more easily predict the commercial success of a product, and

to reduce the costs of gathering information. However, the competition in markets

for standardized goods is usually tough, based upon price, and characterized by

small mark-ups. On the contrary, a new product with a high degree of novelty, such

as in the case of a niche product, increases the perception of product quality by those

consumers in the niche, their willingness to pay, and the firm’s market power.

However, generation of such a product requires detailed information about

consumer needs and the employment of ad hoc technologies, while, at the same time,

increasing the uncertainty of future profits. In relation to this, one should note that,

first, although niche production does not primarily leverage on economies of scale,

but rather on consumer willingness to pay, a minimum threshold of market size is

required if firms want to significantly differentiate their products. Second, the

number of different products introduced into a market also depends on the market

size itself.1

Given the above considerations concerning information about consumers and market

size, the two types of strategy for pursuing variety cannot coexist. Indeed, a strategy

of pursuing novelty, i.e. seeking market power through the specialization in

submarkets that are far from the market standard, necessarily leads to a decrease in

potential market. In the words of Hotelling, the direct effect negatively impinges

upon firms at the extreme limits of the main street. Complementarily, if we believe

1 This result holds both in the address models and in the monopolistic models (for a review see Lancaster, 1990). Guerzoni (2010) analytically proves the validity of this result by considering novelty as the number of product innovations for niches.

5

the theoretical explanation that the production of many products represents a

strategic tool for firms to protect their product portfolio from an unbridled price

competition, then we should rather observe multi-product firms in standard

submarkets, where the competition is higher (and thus where the need to contrast the

price competition is greater) and the market size is large enough to sustain the

existence of many varieties. If this is the case and, therefore, we believe that the

differentiation takes place in standardized markets, we do not expect to observe

substantial modifications of the products in those markets, i.e. changes that are

neither in line with the preferences of standard consumers, nor compatible with a

standardized system of production (Guerzoni, 2010).

In order to empirically tackle this phenomenon, we make innovative use of the

characteristic approach first developed by Lancaster (1966). This approach considers

a product as a bundle of characteristics and the total utility of the product deriving

from accumulation of the utility of each individual characteristic. Saviotti and

Metcalfe (1984) further improved this concept by making a distinction between

service characteristics and technical characteristics, where the former provide a

utility during consumption, while the latter identify the internal (technical) structure,

which allows an artefact to produce services. The characteristic approach has been

widely used to perform hedonic price analysis (Griliches, 1971; Rosen, 1974) to assess

market competitors, and to track technological trajectories (Frenken et al., 1999;

Fontana et al., 2009). Here we use this approach to distinguish between pure novelty

vs. market-related and technology-related differentiation.

As explicitly pointed out by d’Aspremont et al. (1979), the mechanism underlying the

Hotelling strategic effect relies on the consumers’ preferences. Therefore, we make

use of the service characteristics to measure the degree of novelty at the product

level. On the other hand, in order to capture the degree of differentiation at a firm

level, we use both the service and the technical characteristics, investigating their

effects separately.2 Distinctively, in order to capture the concept of novelty, we

2 Please note that the way we formulate our research question and the hypotheses to be tested requires that the two different types of variety are measured at two different levels of analysis: novelty is

6

measure, for each product, its distance from other products of the market in a

multidimensional characteristic space. On the contrary, as a proxy for the degree of

market-related and technology-related differentiation, we use a concentration index

of both technical and service characteristics within the same firm.

Building upon these theoretical and empirical considerations, we examine the

relationship between the introduction of new products in market niches (i.e. the

degree of true novelty in products) and the degree of market-related and technology-

related product differentiation at the firm level. With this aim, we empirically test the

following hypotheses:

H1: the degree of novelty of a new product is negatively correlated with the breadth of the

product portfolio.

H2: the degree of novelty of a new product is negatively correlated with the degree of

differentiation of the product portfolio of the firm that develops it.

H3: the degree of novelty of a new product is negatively correlated with the degree of

differentiation of the production structure of the firm that develops it.

Note that H1 refers to differentiation simply considered as the number of new

products per year. H2 and H3 are required to disentangle the effect of the two costs

of pursuing differentiation: the costs of gathering detailed information about specific

consumers and the opportunity costs associated with ad hoc investments which imply

giving up economies of scale on the production side.

measured at the level of the individual (new) product, while differentiation is measured at the firm level.

7

III. Data and descriptive evidence

Our empirical analysis relies upon an original dataset of innovations including 5109

new skis sold in the European market between 1992 and 2007.3 The main source of

data and information is Sciare, an Italian specialist ski magazine, whose buyers’

guides provide detailed information on key product characteristics. Each year,

companies sell new models in the market, while old skis are usually kept for rental.

This means that for each year our dataset includes an entirely new set of skis. We

collected detailed information on the following variables: price, type of consumer

(beginner, intermediate, expert, professional), style of consumer (e.g. special slalom,

giant slalom, all-round, freestyle) and a set of the technical characteristics of the ski

(core, edges and base materials, anti-vibration system, etc.).

Our sample is made up of 42 firms. On average, we record around 20 firms per year;

however, important differences exist in the sample, since 11 firms produce at least

one product per year (Fischer, Atomic, Rossignol, Salomon, Head, Dynastar,

Blizzard, Völkl, Elan, Dynamic, K2), while 9 appeared in the market for the first time

in 2007 (Sporten, Bottero Ski, Morotto, Duel, Dyad, Nava Ski, AK, Hagan, Hart).

Tyrolia, Authier, Pre, Lacroix and Morotto are present just at the beginning of our

observation period and then exit the market.4 Furthermore, some firms enter and exit

the market more than once – e.g. Kneissl, Volant, Lacroix. Figure 1 shows the number

of different new models introduced between 1992 and 2007, which represents a first

and very rough measure of variety.

{Insert Figure 1 about here}

The total number of new models introduced into the market has substantially

increased over time, from 296 models in 1992 to 552 models in 2007, with a peak of

600 models in 2006. It is interesting to note that this variable was quite stable until

3 Because of some missing data, we perform the analysis on a set of 4202 skis. In some cases, companies’ websites have been used to complement the information available in the magazine. 4 We refer to ‘entry’ and ‘exit’ in terms of new product development. Note that a firm that ‘exits’ the market in our terminology can still be present in the market with old models of skis.

8

1999, had a peak in 2002 (with 510 models) and then decreased substantially until

2006. On average, firms produce 17 new models per year, but there are remarkable

differences over time and across firms. In particular, if we compute the average

number of new models by firm – also taking into account the time of entry/exit in

the market – Fischer, Atomic, Rossignol, Salomon, Head, Dynastar, Blizzard, Völkl,

Nordica and Sporten produce on average 20 or more new models per year. However,

while the first eight firms have always been active in the market, Nordica only

started producing skis in 2001 and Sporten entered the market for the first time in

2007 with 20 different skis. On the other hand, firms like DKB, Olin, Pre and Lacroix

have developed less than six new models per year.

As mentioned before, demand heterogeneity is one of the main sources of variety. If

we segment the market according to consumer skiing preferences, we can investigate

the firms’ patterns of specialization more in depth. To this aim, we group ski models

into 11 overlapping categories, which are highly heterogeneous in terms of technical

characteristics (structure and materials) and target consumers. Table 1 shows the

number of models produced by each company between 1992 and 2007.

{Insert Table 1 about here}

Even when observing market segmentation at the level of different styles of

consumers, some interesting differences emerge across firms (see Table 2).

{Insert Table 2 about here}

First, the market leaders (firms with more than 300 new skis and more than 20 new

skis per year) produce in all ski categories (the only exceptions being Salomon and

Head with 0 products in the alpine segment) and often they are among the top five

producers in terms of number of models produced within a specific segment over

total number of models in that segment.5 Second, the market leaders produce more

5 The alpine and freeride segments constitute important exceptions, since the top producers are respectively Ski Trab and Scott USA.

9

than half of the total models in many categories, but their share of production is

lower for top and alpine skis (around 35 per cent) and for freeride skis (47 per cent),

which are often produced by the niche players. If we examine in more detail the

firms’ strategies, we note that large firms tend to target all the most important market

segments in a similar way. On the contrary, smaller manufacturers tend to focus on

very few market segments: for example, companies like AK, DKB, Dyad Hart and

Olin produce new models either in niche segments such as alpine, or in top market

categories such as racing. All DKB and Hart skis are in this category, and 75 per cent

of Dyad skis are in the freeride niche, but these firms produce nothing in the junior or

easy carve segments.

Analysis of the submarkets at a company level has important implications in terms of

variety. The market leaders produce a very high number of new products and are

present in many different market segments, while firms that target a small number of

market niches also tend to produce a small number of products. Hence, the data in

Table 2 might suggest the existence of a positive relationship between the

development of a high number of new products and the degree of market-related

differentiation at the firm level. However, this descriptive evidence does not provide

any reference to the actual degree of novelty of the new products. In other words, a

firm could be extremely prolific in terms of innovations, by simply inventing around

its existing products, offering very small value in terms of innovative content. On the

other hand, there could be firms that concentrate on a small number of submarkets,

but generate truly new products. In order to further investigate this issue, the next

section presents our empirical analysis, explaining the variables used to test our

hypotheses, the specification of the models and the results.

IV. Empirical analysis

Our empirical analysis aims at understanding the relationship between the degree of

product novelty and the degree of differentiation at the firm level. In order to capture

the degree of novelty, i.e. to what extent a new ski differs from others in the market,

we exploit the information on the skis’ service characteristics and build an indicator

10

of variety with respect to the target market segment for each ski. We proceed in the

following way. We first identify five different service characteristics for each ski,

which refer to the target market: gender/age, carve, top, type of race, style. Each

characteristic can take different ‘values’. In particular, top and carve are either present

(1) or not (0); gender/age can be ‘lady’ (1), ‘junior’ (2) or ‘other’ (0); style identifies

different styles of skiing (e.g. freeride, easy); type of race identifies different types of

race (e.g. giant slalom, special slalom) and can take three different values.6 Thus, each

ski can be represented as a vector of 5 service characteristics and we can obtain 108

different possible value combinations. Each possible combination of service

characteristics constitutes a submarket. If we consider fewer dimensions instead of

considering all 5 service characteristics, the number of submarkets decreases and the

average size grows.

Starting from this, we calculate the dissimilarity index NOVELTY. NOVELTY identifies

the degree of originality of each ski in relation to the other products in the overall

market by year. In order to build this indicator, we first calculate the number of skis

that are identical to the ski under consideration along all the five characteristics

(SIMIL5), i.e. how many skis are in the same submarket when we consider all five

service characteristics. We do the same considering the number of other skis that are

identical along four characteristics (SIMIL4), along three characteristics (SIMIL3), along

two characteristics (SIMIL2), along one characteristic (SIMI1), and along no

characteristic (SIMIL0). Then we create the variable NOVELTYj=

∑=

5

0

*

1

j

SIMILjj

and

standardize it. This variable indicates the degree of novelty of each new ski with

respect to other new skis in the market: the higher NOVELTY, the higher the diversity

of the new ski, i.e. the higher the degree of product originality. Figure 2 shows the

trend in novelty over time (standardized values).

6 Note that this categorization of service characteristics entails a smaller number of categories compared to the one previously discussed at the firm level (see Table 2). This is because some categories are mutually exclusive at the product level (e.g. a ski is either for special slalom races or for giant slalom races) and therefore it is possible to merge some of the categories into a single market segment.

11

{Insert Figure 2 about here}

IV(i) Explanatory variables and their predicted signs

We test our independent variable NOVELTY as a function of three variables of

differentiation at the firm level by year: the number of new products, the degree of

differentiation in technical characteristics and the degree of differentiation in service

characteristics of the firm introducing the product under consideration.

First, in order to investigate the relationship between the degree of novelty and the

breadth of differentiation (our hypothesis H1), we introduce a variable related to the

number of new products by firm. To this aim, we build the variable PRODUCTSHARE,

which is the share of new skis produced by a firm each year and we also control for a

possible non-linear effect, by introducing the square term PRODUCTSHARE2. Following

our theoretical discussion, we expect PRODUCTSHARE to be positively associated with

NOVELTY up to a certain threshold and then to become negatively associated with

NOVELTY, which should be reflected in a negative coefficient of PRODUCTSHARE2. This

is because we expect a higher degree of product divergence to occur in firms that do

not produce a high number of products.

Second, in order to test our hypotheses H2 and H3, we build two different measures

of differentiation, SERVDIFF and TECHDIFF. SERVDIFF is a variable representing the

within-firm differentiation level in terms of market segments. We have mentioned

before that the market can be segmented with the use of service characteristics and

that some firms produce skis for all segments, while others focus more on specific

niches. We can consider the range of market segments served by each firm as another

indicator of within-firm product differentiation. Therefore, we first build a dummy

variable for each market segment, which takes the value 1 if a specific ski is

developed for that segment and 0 otherwise. We then calculate, at the firm level, the

average normalized Hirschmann-Herfindahl (HH) index of market segments by

year, which indicates the degree of concentration related to the range of markets

12

served by each firm. We take marketsHHSERVDIFF −= 1 as an indicator of firm

differentiation in service characteristics. We expect firms that are strongly

differentiated in terms of target markets to develop products with a relatively lower

degree of novelty, as suggested by our hypothesis H2.

The second measure of differentiation, TECHDIFF is built upon three groups of

variables: the core structure of the ski, the edges and the base (see Corrocher and

Guerzoni, 2009, for a detailed description of the database). These variables take value

1 if a specific technical characteristic is present7 and 0 otherwise. For each ski, we first

build the variable STRUCTURE, which is the horizontal sum of the three dummy

variables. Then, we calculate, at the firm level, the average normalized Hirschmann-

Herfindahl index by year in the STRUCTURE, which indicates the degree of similarity

of the technical characteristics across the skis produced by each firm. Since we are

interested in the degree of differentiation, we compute the variable

structureHHTECHDIFF −= 1 where structureHH is the Hirschmann-Herfindahl of STRUCTURE.

We expect firms that are highly differentiated from a technical point of view to

introduce new products with a lower degree of novelty, which is in line with our

hypothesis on the inverse relationship between a product’s originality and the firm

differentiation in terms of technical characteristics.

Finally, we include some control variables, in particular the price of each product, to

control for perceived product quality, firm specific characteristics (through firm

dummy variables), and time trend. The independent variables of the econometric

model are presented in Table 3.

{Insert Table 3 about here}

IV(ii)Results

7 As far as the core structure is concerned, we identify four characteristics: Sandwich, CAP, Torsion Box and Monoblock. The ski edges differ in terms of materials (e.g. iron, steel, diamond) and structure (e.g. trapezoidal vs. segmented edges) and may have different combinations of materials and structures. Finally, we can distinguish between polyethylene bases and graphite bases.

13

Our results are illustrated in Table 4. We first run an OLS regression, the results of

which are reported as Model 1 in the table. However, if we investigate the behaviour

of NOVELTY, we observe that its distribution is far from being normal, which signals

serious problems of heteroskedasticity. For this reason, we investigate the robustness

of our results with alternative specifications. In particular, we run three quantile

regressions. In Model 2 we estimate coefficient by minimizing the absolute distance

from the median, instead of the mean. Because the heteroskedasticity is partly due to

the existence of heavy outliers and the median is less sensitive to them, this model is

more robust than a standard OLS. In Models 3 and 4, we estimate the quantile

regressions for the 25th and the 75th quantile. Almost all coefficients remain stable and

significant across the different specifications, but in Model 2 and 3 the adjusted R2

improves. In all models we also control for firm-specific dummy variables.

{Insert Table 4 about here}

As Table 4 shows, the empirical evidence seems to support our theoretical

hypotheses. To begin with, our first hypothesis – i.e. product diversity is negatively

correlated with the breadth of product portfolio - is partially confirmed. We first try

to use PRODUCTSHARE and we observe a positive sign of the coefficient, but when

controlling for a possible non-linear effect, we find a very high (negative) coefficient

of the square term PRODUCTSHARE2, indicating more complex effects. Since NOVELTY

represents the degree of diversity across products, this result shows that the firms

introducing many new products per year are less likely than others to develop truly

new products (i.e. products that are very different from the market average).

The most important results of our analysis however refer to the relationship between

the degree of novelty and the market- and technology-related differentiation.

Hypotheses 2 and 3 are fully confirmed by the results of our estimates. The

development of innovations (i.e. the development of original products) does not

seem to be compatible with the differentiation strategies either at the level of

technical characteristics (production structure) or at the level of service characteristics

(target markets). This is shown by the fact that both TECHDIFF and SERVDIFF have a

14

negative and significant impact on NOVELTY. Since TECHDIFF and SERVDIFF are

measures of differentiation in technical and service characteristics at the firm level,

i.e. across a single firm’s products. This result implies that a high level of

differentiation is associated with a low level of product novelty. Furthermore, note

that the coefficient of SERVDIFF is higher than the coefficient of TECHDIFF. This result

suggests that the source of conflict between producing novelty and pursuing

differentiation strategies lies in the difficulties of segmenting the market and

satisfying different consumer preferences, more than in the development of ad-hoc

production structures.

In all the models, we control for the effect of price, which always shows a positive

and significant coefficient. This is not surprising, as it is exactly the prediction of the

Hotelling strategic effect: firms developing new products that are very similar to

others already existing in the market cannot exploit the price differentials. However,

here, we do not introduce this variable to test a trivial hypothesis, but rather to

control for quality. In this sense, our results can be interpreted as providing evidence

of the positive relationship between product novelty and quality. We also control for

the effect of time in our analysis by using the variable TREND, which is a time index.

The coefficient of TREND is significant and negative in all the specifications, which has

interesting implications for our analysis of variety. The descriptive evidence on the

sector shows that several new models are introduced in the market every year.

However, our empirical investigation seems to suggest that the variety is

increasingly due to the proliferation of similar products rather than to truly radical

innovations, which means that companies are trying to leverage upon differentiation

strategies to exploit advantages of price discrimination rather than producing

something which is really new and could increase consumers’ benefits.

Generally speaking, these results highlight that the ski manufacturing sector has a

composite structure. On the one hand, we have large multi-product firms that each

year introduce many innovations, but tend to strategically ‘invent around’ their

existing products more than developing truly different products. On the other hand,

we observe medium-sized firms seeking the introduction of truly new products. The

15

former strategy seems to be evolutionarily successful, because over time there are

always less and less radical product innovations, although the number of total new

products per year remains very high.

We can take an important lesson from the analysis of this market. Here we observe a

dual structure, where a large differentiated mainstream market leaves the task to

smaller firms to both serve the market niches and introduce variety. This variety is

progressively incorporated into the development of products for the mass market

over time (see Piore and Sabel (1984) and Guerzoni (2010) about the relationship

between the dualist structure of the industry and the generation of variety). This type

of production organization at an industry level is probably the most efficient way for

mature industries to solve the trade-off between standardization and variety.

V. Conclusions

The aim of the present paper has been to examine the role of variety in a consumer

goods sector (the ski manufacturing industry) and its relevance in firms’ strategies.

In particular, it investigates and empirically tests a series of hypotheses concerning

the relationship between the introduction of truly new products and the

development of small innovations around more standardized products. In line with

the literature, we claim that the former is the true engine of growth and the real

source of industrial dynamics, and that it may not be compatible with the latter,

which simply identifies firm-specific strategies of inventing around existing products

(both in terms of technical characteristics and in terms of target submarkets) for the

purpose of exploiting price differentials.

In particular, we suggest that strategic differentiation at a firm level may hinder the

development of true innovations. In order to test these hypotheses, we have collected

data on products and firms in the ski manufacturing industry, which represents an

interesting case, since it is a mature industry where submarkets can be easily

identified, but where, nevertheless, there is a good deal of differentiation across firms

and diversity across products. As a further contribution, the paper also makes novel

16

use of the characteristics approach, which is here employed to distinguish two types

of variety – the degree of novelty and the degree of technical and market-related

differentiation – and to understand their relationship.

The results of our analysis are encouraging. In particular, we show that, controlling

for quality, highly innovative products are rarely introduced by multi-product firms

or by firms that have a differentiated product portfolio both in terms of target

markets and in terms of technical characteristics. Furthermore, we find that over time

the degree of novelty decreases, suggesting that firms are progressively searching for

variety by marginally differentiating their products instead of developing truly new

products. There are relevant implications of our results both at the firm level and at

the industry level. At the firm level, our results suggest an implication in terms of

market positioning and the pursuit of specialization. Large firms should engage in

competition with niche players only after careful analysis of the competitors’ product

portfolios and vice versa. At the industry level, we can state that although large firms

carry out more R&D activity and develop more innovations over time compared to

smaller market players, they do not necessarily introduce the most radical

innovations, which are the ultimate engine of growth. Finally, the niche market

competitive environment does not necessarily consist of micro firms. Indeed, most

innovative firms are medium-sized companies, because a minimum threshold of

market size is required to introduce innovations, even when a firm does not

primarily leverage on economies of scale.

References

Corrocher, N. and Guerzoni, M., 2009, ‘Product Variety and Price Strategy in the

Ski Manufacturing Industry’, Journal of Evolutionary Economics, 4, pp. 471-486.

David, P.A., 1994, ‘Standardization Policies for Network Technologies— The Flux

between Freedom and Order Revisited’, in Hawkins, R., Mansell, R. and Skea, J.

(eds), The Politics and Economics of Standards in Natural and Technical Environments

(Edward Elgar, Cheltenham).

17

David, P.A. and Rothwell, G.S., 1996, ‘Standardization, Diversity and Learning:

Strategies for the Coevolution of Technology and Industrial Capacity’, International

Journal of Industrial Organization, 14(2), pp. 181-201.

d'Aspremont, C., Gabszewicz, J.J. and Thisse, J.F., 1979, ‘On Hotelling's "Stability

in Competition”’, Econometrica, 47(5), pp. 1145-1150.

Dixit, A. and Stiglitz, J., 1977, ‘Monopolistic Competition and Optimum Product

Diversity’, American Economic Review, 67(3), pp. 297-308.

Fontana, R., Nuvolari, A. and Saviotti, P.P., 2009, ‘Introduction to the Journal of

Evolutionary Economics special issue: The Product Characteristics Approach to

Innovation Studies’, Journal of Evolutionary Economics, 19(4), pp. 463-469.

Frenken, K., Saviotti, P.P. and Trommetter, M., 1999, ‘Variety and Niche Creation

in Aircraft, Helicopters, Motorcycles and Microcomputers’, Research Policy, 28(5), pp.

469–488.

Giraud-Héraud, E., Hammoudi, H. and Mokrane, M., 2003, ’Multiproduct Firm

Behaviour in a Differentiated Market’, Canadian Journal of Economics, 36, pp. 41–61.

Griliches, Z., 1971, ‘Hedonic Price Indexes for Automobiles: An Econometric

Analysis of Quality Change’, in Griliches, Z. (ed.) Price Indexes and Quality Change:

Studies in New Methods and Measurement, (Harvard University Press, Cambridge,

MA).

Guerzoni, M., 2010, ‘The Impact of Market Size and Users’ Sophistication on

Innovation: The Patterns of Demand’, Economics of Innovation and New Technology,

19(1), pp. 113–126.

Hotelling, H., 1929, ‘Stability in Competition’, Economic Journal, 39, pp. 41-57.

Klemperer, Paul, 1992, ‘Competition When Consumers Have Switching Costs: An

Overview’, CEPR Discussion Papers 704.

Lancaster, K.J., 1966, ’A New Approach to Consumer Theory’, Journal of Political

Economy, 4, pp. 132-157.

Lancaster, K.J., 1990, ‘The Economics of Product Variety: A Survey’, Marketing

Science, 9(3), pp. 189-206.

Piore, M.J. and Sabel, C.F., 1984, The Second Industrial Divide: Possibilities for

Prosperity (Basic Books, New York).

Rae, J. B., 1985, The American Automobile Industry (Twayne, Boston).

18

Rosen, S., 1974, ‘Hedonic Prices and Implicit Markets: Product Differentiation in

Pure Competition’, Journal of Political Economy, 82(1), pp. 34-55.

Saviotti, P.P., 1991, ‘The Role of Variety in Economic and Technological

Development’, in Saviotti, P.P. and Metcalfe, J.S., Evolutionary Theories of Economic and

Technological Change (Harwood Academic Publisher, Chur).

Saviotti, P.P., 1994, ‘Variety, Economic and Technological Development’, in Van

Raan, A.F.J. (ed.), Handbook of Quantitative Studies of Science and Technology (North-

Holland, Amsterdam).

Saviotti, P.P., 2001, ‘Special Issue: Variety, Growth and Demand’, Journal of

Evolutionary Economics, 11(1), pp. 119-142.

Saviotti, P.P. and Metcalfe, J.S., 1984, ‘A Theoretical Approach to the Construction

of Technological Output Indicators’, Research Policy, 13, pp. 141–151.

Swann, G.M.P., 2001, ‘The Demand for Distinction and the Evolution of the

Prestige Car’”, Journal of Evolutionary Economics, 11(1), pp. 59-75.

Vassilakis, S., 1992, ‘Multiproduct Firms in the Representative Consumer Model of

Product Differentiation’, Journal of Mathematical Economics, 22, pp. 149-160.

Weitzman, M.L., 1992, ‘On Diversity’, Quarterly Journal of Economics, 107(2), pp.

363–405.

19

Tables and Figures

Table 1 – Industrial dynamics in the ski manufacturing sector, 1992-2007

1992 1993 1994 1995 1996 1997 1998 1999 2001 2002 2003 2004 2005 2006 2007 TOTAL

Pre-existing companies 20 18 20 20 19 15 16 18 15 17 15 15 17 18 19

New entries 0 2 0 1 1 0 1 1 3 3 0 0 1 5 4 22

Old entries* 0 0 1 0 0 0 4 0 0 0 0 4 0 1 1 11

Exits 0 2 0 1 2 5 0 1 3 2 5 0 0 0 4 25

Old exits* 0 0 0 0 0 0 0 1 1 0 0 0 2 0 1 5

*Firms that have entered/exited the market not for the first time.

20

Table 2 – Skis by company and style of consumers Company Race Giant

Slalom Junior Lady Allround Special

Slalom Carve Top Alpine Easy Freeride

AK 2 4 0 0 0 2 2 0 0 0 0

Atomic 101 43 43 48 187 43 87 3 3 31 33

Authier 0 6 8 8 21 7 3 1 2 0 0

Blade 6 8 0 0 2 4 4 0 0 0 4

Blizzard 117 65 38 35 125 49 83 1 3 21 48

Bottero Ski 2 4 0 4 6 2 2 2 0 0 0

DKB 4 4 0 0 0 0 4 0 0 0 0

Duel 4 2 0 2 2 4 2 0 0 0 2

Dyad 6 0 0 2 2 0 0 0 0 0 6

Dynamic 99 25 35 25 106 28 75 2 3 45 23

Dynastar 117 39 32 53 133 32 72 7 8 38 43

Elan 113 40 23 33 111 28 94 4 1 40 37

Fischer 128 59 29 39 190 43 138 1 5 53 11

Hagan 0 2 0 2 3 1 0 0 0 0 0

Hart 6 0 0 0 0 0 4 0 0 0 4

Head 101 34 31 52 162 38 90 1 0 30 32

K2 85 40 15 50 99 37 72 3 3 13 29

Kästle 39 25 40 21 66 24 31 1 7 7 9

Kneissl 50 26 10 33 81 18 41 2 5 8 23

Lacroix 3 4 0 7 17 4 2 0 0 0 1

Longoni 7 1 0 0 4 3 11 0 0 5 0

Maxel 10 10 0 8 22 6 7 0 0 0 3

Morotto 0 2 2 2 4 0 0 1 3 0 0

Nava Ski 2 2 0 2 2 2 0 0 0 0 2

Nordica 66 18 0 23 62 17 54 0 0 22 24

Olin 2 2 0 0 5 1 1 0 0 0 0

Pre 0 3 0 6 8 2 2 1 0 0 0

Prime 12 6 4 0 2 0 14 0 0 7 3

Quechua 13 3 0 6 9 3 10 0 0 5 3

Rossignol 126 37 37 44 146 35 96 2 2 50 41

Salomon 119 37 8 56 171 29 93 4 0 37 38

Scott USA 81 4 1 7 30 12 38 0 0 4 62

Ski Trab 18 25 24 14 35 15 4 4 10 0 0

Spalding 10 16 20 10 37 12 4 2 3 2 0

Sport Specialist 6 6 0 2 2 10 6 0 0 0 0

Sporten 2 4 0 6 12 4 2 0 0 0 0

Stöckli 78 20 6 15 47 17 61 0 0 12 47

Tecno Pro 16 0 0 12 29 3 15 0 0 10 3

Tua Ski 37 8 10 13 62 8 29 2 9 17 13

Tyrolia 9 11 15 13 48 11 3 0 0 3 0

Volant 13 3 2 28 45 4 15 1 0 0 10

Völkl 115 48 21 37 116 41 87 1 4 38 36

TOTAL 1725 696 454 718 2211 599 1358 46 71 498 590

21

Table 3 – The explanatory variables Variable Description

TECHDIFF (1-Hirschmann-Herfindahl indexSTRUCTURE) SERVDIFF (1-Hirschmann-Herfindahl indexMARKETS) PRODSHARELOW Dummy variable indicating if the share of new products is < 0.025 PRODSHAREHIGH Dummy variable indicating if the share of new products is > 0.075 logPRICE Log of the product price TREND Time trend

Table 4 – Regression results

Dependent variable:: NOVELTY

Model 1 (OLS)

Model 2 (QUANTILE 50)

Model 3 (QUANTILE 25)

Model 4 (QUANTILE 75)

SERVDIFF -0.00369*** -0.00438*** -0.00299*** -0.00365***

(0.00037) (0.00024) (0.00017) (0.00032)

TECHDIFF -0.00306*** -0.00426*** -0.00104** -0.00228***

(0.0011) (0.00064) (0.00046) (0.00084)

PRODUCTSHARE 0.0455*** 0.0316*** 0.0207*** 0.0387***

(0.014) (0.0082) (0.0060) (0.011)

PRODUCTSHARE2 -0.334*** -0.250*** -0.238*** -0.211**

(0.10) (0.062) (0.045) (0.083)

logPRICE 0.000300*** 0.000504*** 0.000255*** 0.000636***

(0.000071) (0.000049) (0.000032) (0.000070)

TIME TREND -0.000146*** -0.000139*** -0.000119*** -0.000167***

(0.0000066) (0.0000058) (3.63e-06) (0.0000078)

CONSTANT -0.991*** -0.991*** -0.994*** -0.993***

(0.0012) (0.00095) (0.00057) (0.0011)

FIRM DUMMY VARIABLES YES YES YES YES

Adj R Squared 0.20

Psuedo R Square 0.22 0.25 0.15

Observations 4202 4202 4202 4202

Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

22

0

100

200

300

400

500

600

700

1992 1993 1994 1995 1996 1997 1998 1999 2001 2002 2003 2004 2005 2006 2007

Figure 1 – Number of models per year

-.998

-.997

-.996

-.995

-.994

NO

VE

LT

Y

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

year

Figure 2 – Degree of novelty over time