INNOVATION, STRUCTURAL CHANGE AND DEMAND EVOLUTION: DOES DEMAND SATURATE?*

Pier-Paolo Saviotti, Faculty of Geosciences, Cross-faculty “Institutions” program, sub-theme “innovation”

Utrecht University, Utrecht, The Netherlands and

GREDEG CNRS, 250 Rue Albert Einstein, Sophia Antipolis, FranceE-mail: pier-paolo.saviotti@wanadoo.fr

and

Andreas Pyka, University of Hohenheim, Economics Institute, Wollgrasweg 23, D-70599 Stuttgart,

Germany, E-Mail: a.pyka@uni-hohenheim.de

Published in : J Evol Econ DOI 10.1007/s00191-015-0428-2

Abstract: In this paper the exploration of the interaction mechanisms between the evolution of demand and the overall process of differentiation of the economic system is pursued in a twofold way: (i) the question is asked whether demand indeed saturates and whether such saturation is required for an ongoing economic development; (ii) the impact of the income distribution on the evolution of demand and thus on economic development is investigated. To study the existence and impact of demand saturation, Engel curves are calculated under different conditions. To study the impact of income distribution, TEVECON distinguishes within the population of the economic system two social classes, which differ in their level and quality of education, and thus for their human capital and wages.

Keywords Demand . Innovation . Saturation . Income distribution . Engel curves

JEL Classification O O3 O15 E20

1. INTRODUCTION

In this paper we study the impact of the saturation of sectoral demand on the overall pattern of

economic development. The importance of the problem arise for us, starting from our initial

observation that economic development, at least since the industrial revolution, and in particular

since the beginning of the 20th century, had been accompanied by a growing differentiation of the

economic system. We are convinced that this differentiation is not an epiphenomenon of economic

development but one of its fundamental features. In particular we start finding a possible

* The research leading to these results has received funding from the European Union Seventh Framework Programme FP7/2007-2013 under grant agreement n° SSH-CT-2010-266959; project PICK-ME.

1

explanation for the relationship between differentiation and development in Pasinetti's (1981,

1993) proposed imbalance between saturating demand and continuously increasing productive

efficiency. Such imbalance and the consequent bottleneck in development could be overcome by

the innovative emergence of new sectors, which would increase the differentiation of the

economic system. We developed our TEVECON model of economic development on the basis of

this intuition (Saviotti, Pyka, 2004) and focus our attention on the role of differentiation in

economic development. Hence demand saturation plays a central role in the construction of our

model. Many modifications enhancing the interactivity of different components of the economic

system were introduced in subsequent versions of TEVECON. In particular, in a recent paper

(Saviotti, Pyka, 2013) we start investigating the co-evolution of innovation and demand, arguing

that innovation could not have had any impact on economic development, unless demand and

markets are created for the types of goods and services that it helped to produce. In that paper we

explore the transition from a state of the economic system, in which most people could only afford

to purchase the basic necessities of life to one in which most people can purchase a much wider

range of goods and services of increasing quality and of intra sector differentiation. In this paper

we pursue the exploration of the mechanisms, whereby the evolution of demand interacts with the

overall process of differentiation of the economic system by doing two things: first, we ask the

question of whether demand indeed saturates and whether such saturation is required for

economic development to continue; second, we study the impact of income distribution on the

evolution of demand and thus on economic development. To study the existence and impact of

demand saturation, we calculate Engel curves under different conditions. To study the impact of

income distribution, we distinguish in this version of TEVECON within the population of our

economic system two social classes, which differ for their level and quality of education, and thus

for their human capital and wages.

In what follows we will place our paper in the context of the existing literature. In the past, most

growth models, whether of the exogenous (Solow, 1956, 1957) or of the endogenous (Romer,

1990; Aghion, Howitt, 1992) vintages, did not take demand into account. It is only recently, that

some growth models started including demand in their analysis (Matsuyama, 2002; Foellmi,

Zweimuller, 2006). These models differ from TEVECON in a number of aspects and we will

comment on these differences later. Furthermore, some recent papers started studying the effect

of product quality on demand. For example, Bils and Klenow (2001) and Manig and Moneta (2014)

distinguished between quality and quantity Engel curves. Economic development in TEVECON

occurs by means of three trajectories, which involve (i) a growing number of sectors, each based

on a different type of output, and (ii) a growing output quality and (iii) intra-sector differentiation.

In comparing our paper to the literature, we describe how we define output quality and how that

2

affects the Engel curves we calculate.

3

2. THE EVOLUTION OF CONSUMPTION

2.1 Past and expected trends

One of the most important trends in consumption in developed countries (DCs) during the 20th

century has been the possibility for large sections of their populations to move beyond the

consumption of basic necessities towards a wider range of higher quality and more differentiated

goods and services (Matsuyama, 2002; Foellmi, Zweimuller 2006; Saviotti, Pyka , 2013). Before

that time, in most societies the consumption of the majority of people was limited to basic

necessities of biological origin, including food, clothing and housing. Such pattern of consumption

did not differ substantially from that of other animal species except for the much greater division

of labour through which such biological necessities were supplied in human societies. Although

more sophisticated patterns of consumption had long been present amongst the few rich people,

for most people a more differentiated consumption including goods and services of higher equality

and of greater intra-sector differentiation had to wait until the 20th century, in what we called the

transition from necessities to imaginary worlds (Saviotti, Pyka, 2013). Starting from this transition,

the share of biological necessities in consumption started falling and the share of artificial,

manmade goods and services started rising. Today biological necessities represent no more than

half of overall consumption.

A growing consumption variety has been accompanied by a growing quality and intra-sector

differentiation of all goods and services consumed, including biological necessities. The increasing

quality and differentiation of the food and clothing, that consumers in industrialized countries

consume today is incomparably greater than that consumed in the 19th century. As far as

biological necessities are concerned, the differentiation of their consumption occurred by

introducing more complex tools (Witt, 2001), which could be considered the capital goods of the

household, ranging from beds to storage devices to cooking utensils, giving rise to what we call

'extended necessities'.

What we call imaginary worlds is in fact a heterogeneous category, including objects such as cars,

computers, telephones, leisure activities, education etc. Although such goods and services are not

biological necessities, they become necessary for survival in manmade environments. Not to have

access to internet can be very debilitating today in some advanced societies. Essentially imaginary

worlds correspond to manmade wants generated by the process of economic development. For

example, the growing participation of women in the labour market benefited from the presence of

household capital goods which could speed up domestic work.

The time allocated to leisure activities has generally increased since the middle of the 20th

century. The beginning of paid holidays gave a considerable impulse to consumption and to the

4

economic activities which provide goods and services complementary to leisure activities. Such

activities are unevenly distributed amongst different countries, even at an equivalent level of

economic development. For example, the USA tend to have much shorter holidays than European

countries (Zilibotti, 2010; Stiglitz, 2010). Such trend in consumption was accompanied by a

considerable reduction in working time, which fell from about 70 hour weekly to about 40 hours

weekly during the 20th century (Bouvier, Diallo, 2014; Whaples, 2001).

An even greater reduction in working times had been predicted by Keynes (1928) who expected

that if previous trends had continued most people in DCs would have been able to solve their

economic problem, with which he indicated the consumption of necessities, by working about

fifteen hours per week. In the remaining time they would have been able to acquire the

consumption habits of the middle classes by indulging in 'higher' activities, such as the arts.

Keynes' excessive prediction of the fall in working times, which seems less excessive if we take

into account the length of holidays in some European countries, was due to the underestimation of

the future possibilities of consumption. The large diffusion of consumer durables, of means of

transport and of communication which occurred throughout the 20th century and which

continues in the 21st absorbed more labour time than predicted by Keynes and reduced the

amount of time which could be allocated to leisure activities. Essentially Keynes did not imagine

the differentiation of the economic system and the consequent expansion of consumption which

would occur later.

Summarizing all these trends we can say, that the general development of consumption in

developed countries during the 20th century corresponds to a movement away from biological

necessities and towards manmade goods and services.

As a consequence we can distinguish four categories of consumption:

1. Biological Necessities

2. Extended biological Necessities

3. Manmade structurally induced wants

4. Leisure activities

If we were today to ask the question 'Will the above trends continue in future and affect all the

countries of the world?' we would have to take into account two factors, which were not

considered in any of the previous papers:

First, the previous papers were focused exclusively on DCs. In LDCs the transition from

necessities to imaginary worlds has for long been limited to the richest classes and it has

been largely or completely absent in the general population.

Second, a factor that could severely limit the continuation of the above trends in

5

consumption, not to mention its possible expansion, is the environmental impact of the

corresponding human activities. This constitutes a sudden reminder of the biological roots

of human beings.

In fact, the two above factors seem to be working in mutually contradictory directions. While the

caching up of LDCs, which would imply the adoption of consumption patterns similar to the past

ones of DCs, would be highly socially desirable, it would aggravate the environmental situation of

the earth, unless it were compensated by a sufficient reduction in the environmental impact of the

required technologies and human activities. To achieve an adequate trade-off between growth

prospects and environmental impact is one of the great challenges of this century which is likely to

require important changes in future consumption patterns.

The scope of these of these trends and of the questions they raise is immense and it is quite

unlikely that any individual model can give us answers to all of them. Our TEVECON model is no

exception and we are going to describe its implications and limitations later. We start here by

referring to the way we represent products, services and quality. Products are not purchased for

their material nature but for the services they perform for their users and consumers. We

represent products by means of two sets of characteristics, corresponding to the internal

structure of the product and to the services it performs (Saviotti, Metcalfe, 1984). This

representation can be considered a variant of Lancaster's (1966) characteristics representation.

In turn, the services supplied by a product correspond to consumers' and users' needs and

wants. In this context the difference between material products and 'pure' services is that the

former supply services are embodied in the internal structure while the latter supply services

are of a disembodied form. Furthermore, output quality is measured by the range and by the

level of the services supplied. In terms of this representation, and discarding the possibility of a

return to the consumption of only necessities, the achievement of an adequate trade-off between

growth and sustainability requires the maintenance of the present level of services combined

with a drastic compensating reduction of their environmental impact.

2.2. Innovation, demand and economic development There is now a large consensus about the role which has been played by innovation in the

growth process occurred since the industrial revolution. However, innovation on its own could

never have had an impact on growth, if the goods and services it produced had not been

purchased by consumers and users or, in other words, if there had not been a demand for the

new goods and services. The emergence of such a demand required (i) a disposable income with

which consumers could purchase these goods and services, and (ii) a set of preferences which

made these goods and services desirable.

6

It seems then clear that a theory of demand needs to address the change in the patterns of

consumption which occurred during economic evolution. Although they have not been included

in the textbook theory of demand, some developments in economic theory tried to address the

problem of the change in the content of consumption. A hierarchical theory of wants has been

discussed by several amongst the leading neo-classical economists, such as Walras (1896, 1988),

Jevons (1924), Marshall (1949), but it has received its most explicit and detailed treatment in

the work of Menger (1950), (Saviotti, 2001). Outside economics a hierarchy of needs has been

proposed by Maslow (1943).

2.2.1 Hierarchical theories of wants The terms demand and consumption are not synonyms. Demand can be expected to refer to

wants and needs while consumption refers to the use of goods and services to satisfy wants and

needs. Consumption and its variations in the course of time can be observed while wants and

needs have to be inferred from these observations. A theory of demand and of its evolution

should start from empirical observations of consumption, in order to infer what needs and

wants consumers had at given times and how these needs and wants changed. The theory of

demand which we find in economics textbooks simply says that consumers maximize their

utility starting from a set of preferences which can be assumed as given and constant. Starting

from such a definition the theory of demand can only be a short run one, unless one assumes

that preferences not only are given but have been constant all throughout human history.

Empirical observations show that consumption changed enormously in the course of time by

integrating completely new types of goods and services. The question then arises of whether

consumers could have preferences for objects of consumption of which they knew neither the

existence nor the properties. There is a growing acceptance that preferences can be endogenous

(Aversi et al. 1999; Stiglitz, 2008) and that they can evolve in the course of time. We do not

attempt to give a systematic answer to this question except for saying that consumers cannot be

expected to have precise preferences for objects of consumption unknown to them. At best we

can assume consumers to have broad and imprecise preferences for particular types of

'services', such as transport, communication, leisure and need to learn how new objects of

consumption could supply these services. The learning process is not be limited to consumers

but needs to involve innovators and entrepreneurs. Following Schumpeter (1912, 1934), we

expect entrepreneurs to detect in a fuzzy form emergent wants, to create goods services to

satisfy them and to teach consumers about their use.

Following Witt (2001) we discuss the evolution of consumption as an evolution of needs and

wants. In this sense, necessities are rooted in biological needs, even if the different ways in

which these necessities can be satisfied can partly be due to wants generated by social

interactions. Additionally, imaginary worlds can be either wants which do not correspond to any

7

biological necessity but are induced by the evolution of the socio-economic system or leisure

activities.

This interpretation of the evolution of consumption bears some similarity to a hierarchical

theory of consumption (Menger, 1950) according to which necessities are lower goods and

services and imaginary worlds are higher goods and services. Such a categorization is

incomplete since it neglects the internal quality and differentiation of the output of different

sectors. Even goods and services corresponding to biological necessities are produced with

growing levels of quality and differentiation. Furthermore, according to Witt (2001) the

consumption corresponding to biological necessities can be coupled with tools or services which

provide enhanced consumption opportunities. We can consider these complementary types of

consumption 'extended necessities'. In this respect, the time trends of necessities and of

imaginary worlds are the same. In both cases we can expect to observe within existing sectors

growth in output quality and in intra-sector differentiation. Various sectors could differ for the

balance between efficiency growth, which reduces costs and prices, and quality growth, which

raises costs and prices. Thus, in rich countries, the share of household expenditures allocated to

food or clothing tends to fall with growing income. Likewise, the prices of electronic goods tends

to fall simultaneously with considerable increases in output quality and differentiation.

Additionally, in rich countries the price of housing tends to rise as a share of household

expenditure. Thus, one of the necessities starts being out of reach for some members of society.

The above trends imply that:

(i) Neither necessities nor imaginary worlds can be treated as homogeneous categories.

Different components of each of them can have different time trends giving rise to structural

change in consumption

(ii) Even if a hierarchical scale of consumption could be established for a population of

consumers there would be considerable differences amongst the members of such a population.

Furthermore, we could expect the extent of variation to be greater for imaginary worlds than for

necessities

What we call imaginary worlds include some very different types of consumption, ranging from

leisure activities to health care to various types of consumer durables. While none of them are in

a biological sense necessities, some of them become 'necessary' in given societies. For example,

not to be able to drive a car or to use a computer can be a serious handicap in modern societies

substantially limiting peoples' ability to find meaningful occupations. In a different sense, the

diffusion of consumer durables such as fridges or washing machines is likely to have been at

least partly induced by the growing participation of women in the labour market.

8

The main point that can be made here is, that the evolution of consumption that occurred over

the last century cannot be due to the emergence of new goods and services for which there has

always been a latent demand. On the contrary, such latent demand and a corresponding set of

preferences can at best have existed in a fuzzy form for some of the present types of

consumption. Other goods and services could simply have been unimaginable. In general we can

expect preferences for new or modified goods and services to be formed by a process of mutual

learning between producers and consumers. The whole process could be started by a

Schumpeterian entrepreneur having an intuition about emerging wants and to continue with

producers needing educate consumers, as Schumpeter thought, until the once emerging goods

and services become so mature that a wide awareness of their existence and knowledge of their

properties are present in a population of consumers. Unsurprisingly, with the availability of this

growing variety in the supply of goods and services marketing has become an essential tool in

many sectors. Furthermore, consumers need to have a growing level of human capital to be able

to consume many of the new goods and services (Saviotti, 2001).

In an interesting paper, somewhat different from the work that made him famous, Keynes’

(1928) explored what he thought would have been the evolution of the capitalist economic

system in the advanced countries of the time. He expected that, if the same trends in technical

progress experienced in the previous century were to continue, most people not only would

have become wealthy enough to have solved what he called 'the economic problem' but could

have done so by working a falling number of hours per week, for example fifteen hours. In fact,

what he meant by the term 'the economic problem' was the satisfaction of what we call

necessities. In Keynes' view most people could have spent the rest of their time in 'higher'

occupations. These higher occupations in Keynes' view seemed not to have an economic nature

(Boldrin, Levine, 2010). Keynes' predictions seemed to be based on the assumption that the

economic system had a fixed composition, in the sense of being constituted by the fixed number

of sectors required to produce necessities with the extent of division of labour that was then

current. However, even without fully accepting Pasinetti's (1981, 1993) assumption of complete

saturation in incumbent sectors, which we will see needs to be relaxed, we now know that the

economic system has become far more differentiated than at Keynes' time and that this growing

differentiation provided an important growth mechanism. What we called imaginary worlds

correspond to a growing number of sectors which have gradually reduced the share of

necessities in overall output and demand. In this context the prediction that working time would

have considerably declined is problematic for two reasons:

(i) It is not certain that if working time had fallen as Keynes predicted it would have been

possible to keep incomes per capita constant or rising. Our work tells us (Saviotti, Pyka, 2004,

9

2008) that growth requires the emergence of new sectors, something which Keynes did not take

into account.

(ii) The new sectors which emerged after Keynes' paper provided the new employment

opportunities that were required to allow the continuation of growth and of increasing incomes

per capita.

As a consequence the emergence of new sectors limited the possible fall in working time, a fall

which nevertheless would have been very problematic without such a change in the composition

of the economic system. Thus, one of the reasons for which Keynes' prediction did not come

true, at least for what concerns the balance between work and leisure time, is the growing

differentiation of the economic system and the employment possibilities it created. Other

reasons for which Keynes' predictions were not entirely correct are discussed in the book edited

by Pecchi and Piga (2010).

In this paper we interpret the changing composition of wants and of patterns of consumption in

terms of the co-evolution of demand and innovation. The differentiation of the economic system

is due to the combination of three trajectories, each one representing the long term trend

towards the rise of efficiency, variety and quality (Saviotti, Pyka 2013). These trajectories are

not independent. In particular, according to Pasinetti (1981, 1993) the saturation of sectoral

demand can lead to a potential imbalance between continuously increasing productive efficiency

and saturating demand. Such imbalance, which could entail a serious bottleneck for economic

development, could be overcome by the creation of new sectors based on new activities and

outputs. In turn, the emergence of new sectors would raise the variety of the economic system.

Thus, the saturation of demand could contribute to the differentiation of the economic system. It

is then quite important to establish whether demand does indeed saturate. To test the existence

of demand saturation we calculate Engel curves under a range of circumstances represented by

different values of TEVECON parameters.

2.2.2 The TEVECON model We are going to explore the evolution of consumption in terms of a model in which economic

development is mainly due to the differentiation of the economic system by the emergence of

new sectors, called TEVECON. Such a model is intended to analyze economic development over

the long run, for example over a period such as the one from industrial revolution to the present.

TEVECON is both an endogenous growth model and a model in which the number of sectors is

endogenously variable. The latter feature is specific to TEVECON and we do not know of any

other model which does that. One of the objectives of the creation of TEVECON was the need to

connect the macroeconomic level of aggregation with micro and with intermediate levels of

10

aggregation. As a consequence TEVECON has both intra and inter sector dynamics. A sector is

defined as set of firms producing a unique although differentiated output. Each sector is created

by a Schumpeterian entrepreneur exploiting a radical innovation in the expectation of achieving

a temporary monopoly. If the innovation is successful, there is going to be a bandwagon of

imitators which is going to reduce the degree of temporary monopoly and to increase the

intensity of competition, thus gradually eliminating the main inducement to entry. In this

process the net rate of entry gradually falls to zero and the rate of exit starts dominating, leading

to a growing industrial concentration. In fact, this process is better described as the transition

from an initially innovative sector to the 'circular flow', the set of routines of an economic

system. With the saturation of a sector we indicate the gradual loss of differential opportunities,

which induce entrepreneurs to enter. This dynamics is represented by a system of equations (Eq

1):

(1) N it=k1 ∙FA i

t ∙ AG it−IC i

t−MA it

where:

Nit = number of firms in sector i

k1 = parameter affected by a number of factors, including the costs of creating a firm, the

presence of entrepreneurs in the economic system etc

FAit = Financial availability in sector i

AGit = adjustment gap in sector i

ICti = intensity of competition in sector i

MAti = failures, mergers and acquisitions in sector i

The rate of entry into a sector i is then determined by the 'financial availability' and by the

‘adjustment gap’ in the same sector. Here 'financial availability' does not just reflect the quantity

of money in the economic system but also the presence of a financial community able to

understand the economic potential of an innovation. The meaning of the ‘adjustment gap’

variable can be understood by thinking that when a radical innovation is first launched there are

neither the demand nor the production capacity for the intended output of the sector. At any

time the ‘adjustment gap’ can be calculated as the difference between maximum potential

demand and the instant demand in the sector. Both demand and production capacity are

gradually created during the life cycle of the sector. As a consequence for a given sector the

‘adjustment gap’ measures the distance from the saturation of the corresponding market. The

adjustment gap is included in the entry term because initially it corresponds to the potential

11

market size. The rate of exit from sector i can be expected to increase with increasing ‘intensity

of competition’ in sector i (ICt) and with the ‘rate of failures, mergers and acquisitions’ in the

sector (MAti). In each sector in a wide range of conditions the number of firms describes a life

cycle, increasing first until a maximum and then falling to a lower value, corresponding to an

increasing industrial concentration and to the saturation of the sector.

We describe the state in which sectoral demand is saturated and in which the intensity of

competition reaches the level prevailing in the incumbent sectors, which corresponds to the

Schumpeterian circular flow, as a state of 'saturation'. Saturation thus conceived plays a very

important role in TEVECON because it represents the loss of profit and growth opportunities,

which had initially induced entrepreneurs' innovative entry. It is this situation of falling

opportunities which induces entrepreneurs to look for other innovations which can be used to

construct new niches and markets. The role played by generalized saturation in TEVECON

justifies the attention we pay to demand saturation in this paper. In turn, the creation of new

sectors represents the inter-sector dynamics of TEVECON.

The number of sectors in TEVECON changes endogenously as entrepreneurs, after having

created a new innovative sector and having it seen to mature and to lose its profit and growth

potential, start looking for other innovations which could lead to new potential markets. If the

saturation of existing sectors represents the inducement to the creation of new ones, new

innovations must be available to create other sectors. The pool of knowledge from which these

innovations can be generated derives from two types of search activities: (i) Fundamental search

activities (FSE) create basic knowledge which can be used in all sectors and speeds up the rate of

their creation; and (ii) sectoral search activities (SEi), which create knowledge specific to each

sector, affect output quality and differentiation and, indirectly, sectoral demand. If the pool of

knowledge can give rise to new innovations when older sectors saturate, then new sectors can

be created, leading to increasing output variety. Both FSE and SEi are generated by the allocation

of economic resources: FSE is funded as part of general investment while SE i is funded by firms

in the sector.

Within TEVECON demand coevolves with innovation and with other aspects of the economic

system. For instance, sectoral search activities increase with increasing demand but search

activities themselves indirectly affect demand through their impact on product quality and on

product differentiation (Eqs 2, 3, 4).

(2) SEit=SE0+k4 ∙[1−exp (−k 5∙ Dacc

t )]

Where:

SEit = sectoral search activities in sector i

12

Dacct = demand accumulated at time t

k4 = parameter representing the effect of accumulated demand on search activities, which we

call technological opportunity

k5 = parameter representing the effect of the speed with which demand affects search activities

Eq 2 shows that demand affects search activities, and Eqs 3 and 4 (see below) show that demand

is affected by Yi and YΔ i,, which in turn are affected by search activities. A further co-

evolutionary mechanism, present only in Eq 4, occurs by means of the creation of the disposable

income required to purchase the output of a new sector in addition to those of the preexisting

ones. This new disposable income is created by the joint effect of the growing productive

efficiency of older sectors and of the income created by the emergence of the new sectors.

We hypothesized that a further co-evolutionary mechanism, in which product quality, wages

and education participate in a feedback loop, contributes to the transition from the low quality

(LQ) to the high quality (HQ) goods and services that we called transition from necessities to

imaginary worlds (TFSC, 2013).

The macroeconomic level of aggregation is calculated by aggregating firm and sectoral

outcomes. As an example of TEVECON results sectoral and aggregate employment are shown in

Figs 1 and 2. There we can see that even when employment creation falls at the sectoral level an

appropriate succession of emerging sectors of the right size and at the right times can

compensate for the falling capacity of individual sectors to create employment.

1 65 129193257321385449513577641705769833897961 -

50

100

150

200

250

300

350

t

Li

Fig 1. The evolution of sectoral employment Li

13

1 57 113169225281337393449505561617673729785841897953 -

100

200

300

400

500

600

t

# aggregate employment & employment trend

Fig 2. The evolution of aggregate employment and trend of employment

The consequence of this dynamics is that the composition of the economic system needs to

change in long run economic development.

The previous considerations show that the evolution of demand cannot be analyzed in isolation

from that of the rest of the economic system. The evolution of the ‘composition’ of demand, or of

the distribution of demand amongst the different types of goods and services that people

purchase at different times, becomes a relevant task for a theory of demand. Such a theory needs

to be constructed in the context of the evolution of the whole economic system. The analysis of

demand, and in particular of its possible saturation, will constitute the main part of this paper.

Before passing to its description we need to describe some more aspects of TEVECON.

The pattern of economic development observed since the industrial revolution has been

accompanied by a considerable differentiation of the economic system. Growing evidence for

this differentiation can be found in a number of references (Frenken et al ,2007; Funke,

Ruhwedel, 2001a, 2001b, 2005; Hidalgo et al 2007; Imbs, Wacziarg, 2003). This process of

differentiation consists of: (i) the emergence of new industrial sectors based on new economic

activities and on new types of goods and services; (ii) the increasing quality and differentiation

of the output of incumbent sectors.

The observed economic development process occurred can be explained by the following three

trajectories, or long term trends:

TR1: growing productive efficiency,

TR2: growing output variety

TR3: growing product quality and differentiation.

14

TR1 consists of the reduction in the inputs required to produce one unit of a constant type of

output. TR2 consists of the increase in the number of distinguishable output types that the

economic system can produce. TR3 consists of the increase in sectoral output quality and intra-

sector differentiation.

The previous trajectories are not independent but complementary. None of them alone could

have generated or explained the observed pattern of economic development. An increase in

productive efficiency alone (TR1) would have reduced the amount of labour and of the other

inputs required to produce one unit of output. Without the emergence of new sectors this would

have been likely to give rise to growing unemployment. Both growing output variety (TR2) and

growing product quality and differentiation (TR3) can compensate the potential labour

displacing effect of TR1, but none of them could have emerged without the increasing productive

efficiency of TR1. Such increasing efficiency gave rise to a surplus which could be invested in the

search activities required to generate TR2 and TR3. In summary, the combination of these three

trajectories led to a growing output variety of the economic system and consequently to a

growing variety of consumption possibilities.

Within this growing range of consumption possibilities different goods and services play

different roles. Some goods can be considered necessities in a biological sense while others are

manmade (artificial) types of consumption which in a past paper (Saviotti, Pyka, 2013) we called

imaginary worlds. The term imaginary means that these new goods and services are not

required for our biological survival but are an artificial construct of human societies.

TEVECON is a very 'spacious' model containing many variables and having gone through a large

number of modification since it first creation. A description of its general aspects has been given

so far. More specific aspects relevant to this and to recent papers will be discussed in section 3.

Given the space constraints of a normal research article we cannot give description of the model

here. A synthetic table containing the most important equations of TEVECON can be found in the

Research Gate pages of both Pier Paolo Saviotti and Andreas Pyka†.

Compared to the papers by Matsuyama (2002) and by Foellmi and Zweimuller (2006), which are

the most similar to ours, our paper differs both for a general methodological choice and for some

more detailed research objectives. TEVECON is an evolutionary model which does not assume

either perfect knowledge or optimizing rationality. Our economic agents are adaptive improvers,

the distribution of whom varies in the population of each economic system. Furthermore, in

TEVCON we can detect regions of parameter space in which development occurs in an ordered

† https://www.researchgate.net/profile/Pier-Paolo_Saviotti https://www.researchgate.net/profile/Andreas_Pyka

15

way and regions in which becomes chaotic or collapses (Pyka, Saviotti, 2011). Like

Matsuyama(2002) TEVECON is based on a dynamical systems modelling approach, but ours is

not derived from the application of a preconceived set of equations but is inductively

constructed by assembling sets of relationships between variables mediated by parameters and

by introducing within these relationships further connections which gradually increase the

interactivity of the of the TEVECON model. These features of TEVCON give us both advantages

and disadvantages. On the positive side we can construct a model of very considerable scope and

flexibility. We can experiment with a large number of variables and with many inter-variable

interactions thus achieving levels of complexity which, although small with respect to the real

world, are greater than those of more constrained analytical or orthodox models. In a sense

TEVECON is intermediate between orthodox (OM) and agent based models (ABM). It has a

greater modelling flexibility than the OM although smaller than that of ABM. On the other hand,

TEVECON is less ad hoc, more overarching and coherent than ABM models. The disadvantage of

TEVCON with respect to OM models is the absence of closure conditions, such as general

equilibrium, which can provide internal means of logical testing of theories and models. TEVCON

can only be tested by comparing predicted with empirical results. Given the generality of

TEVRCON this mode of testing cannot mean the identity of numerical value of system properties

but is limited to the verisimilitude of calculated and observed trends.

In terms of detailed research objectives we differ from Matsuyama (2002) for the choice of

learning mechanisms. Whereas he uses learning by doing we rely predominantly on search

activities, a general analogous of R&D introduced by Nelson and Winter (1982). In this sense,

our model is closer to Foellmi and Zweimuller (2006). However, we differ from them for the use

we make of income distribution: They study its effect on growth while we focus on the impact of

income heterogeneity on the saturation of demand.

We now proceed to the core of this paper, that is to the study of demand saturation and of the

role it can play in economic development.

3. DOES DEMAND SATURATE?

Complete demand saturation occurs when demand stops increasing and remains constant after

a given level of income is attained. This is the type of saturation hypothesized by Pasinetti

extrapolating the work of Engel (1857). In previous papers, we accepted Pasinetti's idea that the

bottleneck originated by the imbalance between saturating demand and continuously increasing

productive efficiency could have been overcome by the emergence of new sectors giving rise to

16

growing output variety. In this sense demand saturation was an essential component of the

mechanism by which economic growth could continue in the long run. Here we challenge this

assumption and ask the questions: Does sectoral demand indeed saturate? Is this saturation

required to allow the long run continuation of economic growth? To answer these questions we

compare the cases of (i) saturation, (ii) under-saturation and (iii) oversaturation. (i) occurs

when demand initially rises and then becomes constant as income rises, (ii) occurs when

demand rises at a diminishing rate without ever becoming constant as income rises, (iii) occurs

when demand falls as income rises. Recent empirical work on Engel curves (Chai, Moneta, 2010,

2011, 2012) shows that complete demand saturation rarely occurs and that non-saturation is a

more frequent outcome, even if more complex Engel curves are observed in a number of cases.

Furthermore, we study the way demand saturation is affected by income distribution.

Following our objective of exploring the co-evolution of demand and of other components of the

economic system, we do not just try to calculate Engel curves but wish to explore the way they

are affected by several non demand variables and parameters. We do not conceive saturation as

an inherent property of demand at a sectoral level, but as a property which is itself determining

and determined by other non-directly demand related economic phenomena.

To achieve the objective described in the previous paragraph we compare the Engel curves

calculated using different versions of the TEVECON model and with different parameter values

in order to see (a) if demand saturation occurs and (b) what circumstances and model

parameters affect the occurrence and type of demand saturation. First, we show the Engel

curves for a version of TEVECON which included a limited extent of co-evolution of demand and

innovation; second, we introduce a stronger form of co-evolution of demand and innovation

combined with the income distribution originating from the distribution of competencies within

two social classes, called low (L) and high (H), differing for their levels of education and of

human capital.

In the first version of TEVECON (v1) sectoral demand depended on product price, product

quality and differentiation, but it is not affected by disposable income (Eq. 1). In this demand

function there was co-evolution of demand and of search activities, but demand itself was not

dependent on disposable income. In the new versions of TEVECON (v2) the sectoral demand

function contains a disposable income term. The two demand functions are shown in equations

3 and 4.

(3) Dit=Di

0 ∙Y it ∙∆ Y i

t

pit

17

-

1

2

3

4

5

6

- 0.02 0.04 0.06 0.08 d1 d2 d3

Engel Curves (kw = 0.25)

-

1

2

3

4

5

6

- 0.02 0.04 0.06 0.08 0.10 0.12 d1 d2 d3

Engel Curves (kw = 2.0)

(4) Dit=Di

0 ∙DDisp, it ∙

Y it ∙∆ Y i

t

p it

where

Dit = demand for good or service i at time t;

Di0 = demand for output type i at the time of its introduction;

DDisp,i = the disposable income which can be allocated to output type i at time t ;

Yit = the quality of output i at times t;

YΔ it = the range within which output quality varies, which measures output differentiation;

pit = the price of output i at time t.

The sectoral demand in Equation 4 will be zero unless the term Ddisp,i, representing the

disposable income for sector i, is greater than zero and its level will be affected by the value of

disposable income. Thus, the term Ddisp,i, in equation (4) adds a further mechanism of co-

evolution to that relating demand to search activities. In this enhanced form of co-evolution of

demand and innovation a new sector i will emerge only if a non zero demand for its output exists

because such non zero demand requires the existence of an adequate disposable income. In turn,

demand increases with innovation because the terms Yi, YΔ i and pi are affected by search

activities (SEi) and search activities are affected by demand. The sectoral demand in equation (3)

is not limited by disposable income and we can expect it to differ from that of equation (4) by

overstating the demand level and possibly by overstating the effect of product quality and

differentiation on demand.

18

Figs 3a and 3b. Engel curves calculated for the TEVECON version v1 with different values of the wages parameter kw. In the two figures the different thickness of the curves corresponds to different sectors. Demand on the vertical axis, total income on the horizontal axis.

The Engel curves in Figs 3a-3b were calculated using demand function (3) and varying the

values of kw, where kw affects wage levels at equivalent labour productivity. In each of these

figures, Engel curves are calculated for three subsequent sectors. We can immediately see that

saturation defined as zero rate of growth of consumption is not attained. The apparent satiation

which seems to occur for very long times is an artificial effect due to the fact that we stop the

runs of our model after a given time. The Engel curves we observe are a series of steps each

constituted by a period of growth followed by a period of quasi constant consumption. The

beginning of each new step coincides with the emergence of a new sector. Thus, if the number of

sectors in the economic system remained constant, Engel curves could achieve saturation.

However, the emergence of new sectors induces a shift in the satiation level of sectoral

consumption by means of the rise in income that it causes. In other words, the shape of Engel

curves is affected by both the tendency to satiation which would occur at a constant number of

sectors and by the shift in the level of satiation induced by the emergence of new sectors. Thus,

satiation is not an implausible assumption as long as the number of sectors is constant, but

becomes implausible due to the structural change, which occurs naturally and which we

maintain is a determinant of economic development.

It is to be noticed that the slope of the growing parts of Engel's curves changes from positive to

negative as the wage rate rises (Figs 3a and 3b). Thus, the higher the wage rate, the more likely it

is, that older goods and services become inferior as they mature during their life cycle. The

period in which the slope of Engel curve becomes negative coincides with that in which the slope

of the income becomes negative after having risen above the trend. In other words, when income

after having grown very fast in the early phases of the new sector starts falling, older goods and

services start being perceived as inferior relative to the new ones. We also observe that the

maximum consumption level of the step corresponding to the emergence of the sixth sector (the

last in our simulation) is not affected by the change in wage rates.

Another novelty of this paper relative to our previous ones is the inclusion of income

distribution amongst the variables, which can potentially affect demand and demand saturation.

We expect that different social groups or classes can have not only different incomes and

incomes per head but also different dynamics of their relative incomes, which potentially affect

Engel curves and demand saturation. The newest TEVECON version (v2) uses the demand

function in Eq. (2) and includes two social classes, differing for their levels of education and of

human capital, giving rise to a distribution of competencies and to a corresponding income

19

distribution. The L and H classes can be compared to blue and white collar workers, or to low

and high skill workers respectively. We can expect the relative income of the two classes to be

affected by the initial distribution of population between the two classes, by the relative

allocation of educational resources, and by the relative quality of education.

If we start with a social system in which the population share of class L is represented by the

term Spop, we can then vary (i) the investment in education as a share of total investment, (ii) the

allocation of this investment to the two classes H and L, and (iii) the relative quality of education

for the two classes. To simplify the comparison, we limit our analysis to the variation of

education's share of total investment (IEd). As we can see from Fig. (4), to vary the education

share of total investment from 5% to 20% considerably affects the evolution of population

shares (Spop). In both cases investing in education raises the share of the H class and reduces that

of the L class. Furthermore, the rise in the share of the H class is much faster and more

pronounced for the higher (20%) than for the lower (5%) education share of total investment.

2 258 514 770 1026128215381794 -

0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

t

spop

2 258 514 770 1026128215381794 -

0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

t

spop

Fig. 4. Evolution of the population shares (spop) of the two classes L (blue) and H (red) when the

education share of total investment is 5% (Left) or 20% (right). The initial share of the

population in class L is 80% in both cases.

The Engel curves for the two levels of education share of total investment are shown in Figs 5

and 6.

20

- 2 4 6 8 10 12 14 16 18 20 -

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Engel Curves for Low SkilledD

Y - 2 4 6 8 10 12 14 16 18

- 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Engel Curves for High SkilledD

Y

Fig. 5. Engel curves for the L (left) and H (right) classes for a 5% education share of total

investment. Demand on the vertical axis, Income on the horizontal axis.

- 2 4 6 8 10 12 14 -

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Engel Curves for Low SkilledD

Y

- 5 10 15 20 25 30 -

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Engel Curves for High Skilled

Y

D

Fig. 6. Engel curves for the L (left) and H (right) classes for a 20% education share of total

investment. Demand on the vertical axis, Income on the horizontal axis.

The shape of the Engel curves differs substantially between the two social classes. Engel curves

for the L class show oversaturation, in the sense that demand for the output of any sector

gradually falls with rising income. On the other hand, Engel curves for the H class show non-

saturation in the sense that the rate of growth of demand, while remaining positive, falls with

income without becoming constant. We can also notice that the demand level, which sectors

tend to achieve as they move towards maturity, is affected by the education share of total

investment. More precisely, when IEd rises the levels of demand increase for the H class, but fall

21

for the L class. In other words, as investment in education and with it the share of the H class in

population rise, people in the L class have to sacrifice part of their consumption in pre-existing

(incumbent) sectors to be able to consume the output of new and emergent sectors. No similar

sacrifice is required of the people of the H class: They are able to add the consumption of the

output of new sectors while increasing their consumption of the output of incumbent sectors.

The Engel curves for the whole population, combining the consumption of the H and L classes,

show almost demand saturation.

These results can be interpreted as follows:

The saturation of demand hypothesized by Pasinetti does not occur in general. In most of the

situations we explored complete saturation does not occur. The Engel curves calculated with the

v1 version of TEVECON seem to be further away from saturation than those calculated with the

v2 version. This can be explained by the fact that the absence of the disposable income term in

the demand equation (3) can be expected to overstate demand levels, thus leading to excessively

large steps due to the emergence of new sectors. On balance, we expect the Engel curves

calculated with the v2 version of TEVECON and including disposable income to be closer to real

ones.

A relatively common behaviour detected so far in our experiments consists of either under-

saturation or oversaturation, where the former is defined as a progressive fall in the rate of

growth of sectoral demand with income as a sector matures, while the latter is defined as a

progressive rise in the rate of growth of sectoral demand with income as a sector matures. In

both cases the rate of growth of demand with income is very high in the initial phases of the

industry life cycle, corresponding to a situation in which the sector is far from saturation, and

starts falling at a time which seems to correspond to the onset of the saturation of the sector.

The shape of Engel curves differs considerably for the two social classes L and H. For example,

under a wide range of circumstances the Engel curves of the L class show oversaturation while

those of the H class show non-saturation. Thus, demand saturation for the whole population of

an economic system could be an artificial result of the aggregation of individual demands, none

of which saturates. The shape of the aggregate Engel curves depends on the relative size of the

two classes, and on their endowment and quality of education. This implies that although higher

levels of education are an advantage for society as a whole, those members who cannot get

access to education become relatively disadvantaged.

The shape of Engel curves is affected by wage levels and by technological opportunity. This is

logical since both of these TEVECON model parameters affect directly or indirectly demand.

22

Furthermore, the very same process of differentiation underlying economic development can

affect Engel curves. If the number of sectors were fixed and if output could somehow continue to

grow, there would be no need for any reduction in expenditures on existing goods and services

to buy emerging ones.

As we have expected, the shape of Engel curves is not an intrinsic and invariable property of

demand but is affected by the interactions of demand with many other components of the

economic system and by many TEVECON model parameters. This is completely natural when

there is co-evolution of many or most components of the economic system. As a consequence the

dynamics of demand, including its possible saturation, cannot be separated from the dynamics of

the whole economic system. In the co-evolution of innovation and demand, saturation can be

both, a cause and an effect of the emergence of new sectors, contributing to their emergence but

being affected by their presence and dynamics later on.

The previous results have considerable implications for processes of economic development.

According to Pasinetti (1981, 1993) the emergence of new sectors can be induced by the

imbalance between continuously growing productive efficiency and saturating demand.

However, complete saturation of demand is not required. An imbalance can occur even if the

rate of growth of demand is inferior to the rate of growth of productive efficiency. In this case we

can expect the intensity of all the inputs required to satisfy demand to fall over time, thus

entailing a gradual fall in the capacity of the incumbent sectors to create employment. This

induces a shift in investment from incumbent mature sectors to emerging ones. Furthermore, we

have to take into account that an economic system in which search activities are carried out all

the times, can create opportunities which could be exploited by entrepreneurs, irrespective of

the above imbalance between productive efficiency and demand. In other words, while in

Pasinetti's ideas the emergence of new sectors is induced by a form of diminishing returns to the

exploitation of incumbent ones, a further mechanism of economic development are driven by

the opportunities presented by search activities. The two mechanisms are described as threat

based or opportunity based. The former bears a considerable resemblance to the mechanism

which according to Nelson and Winter (1982) leads to most types of innovation: The failure to

achieve existing targets based on present routines induces firms to switch to different routines

using the results of search activities.

To this we add that the concept of demand and market saturation is considerably amplified by

the growing product/output quality and differentiation. This introduces a difference between

volume and value saturation. The former would be attained, when the output of a sector -

measured in volume terms - satisfies a demand, which in this respect is constant. The latter is

23

attained, when the quality and the differentiation of the output of a sector can no longer be

increased. Given that the price of a product is likely to be positively related to its quality and that

product quality and differentiation can be expected to grow together, we can expect the average

price to rise and its distribution to become wider within a given sector in the course of time. The

attainment of value saturation then depends on technology, which reduces unit costs and

increase product quality and differentiation, and on disposable income. As a consequence of the

previous considerations a sector, which has attained volume saturation, has not necessarily

attained value saturation. In general, we can expect value saturation to be attained after, and

possibly long after, volume saturation.

The introduction of product quality and differentiation modifies the possibility of compensation

for the imbalance identified by Pasinetti. Whereas in Pasinetti compensation for the imbalance

between productive efficiency and demand comes from the emergence of new sectors, and thus

from growing output variety, in our TEVECON model, compensation comes from both (i) the

emergence of new sectors and from (ii) the growing product quality and differentiation within

each of them. This leads to the possibility of multiple development paths arising from different

combinations of (i) and (ii). Furthermore, the development paths based on different

combinations of (i) and (ii) entail substantial differences in a number of variables such as wages

and the quality of human capital, and in macroeconomic outcomes such as the rates of growth of

income and employment (Saviotti, Pyka, 2013).

4. DEMAND CURVES, OUTPUT QUALITY AND CONSPICUOUS CONSUMPTION

The trend towards rising product quality and differentiation is also likely to affect demand

curves. Such growing quality and differentiation can only occur if consumers have a disposable

income allowing them to purchase the corresponding products. In a previous paper (Saviotti,

Pyka, 2013) we have shown that such disposable income can be created by the combination of

growing productive efficiency in incumbent sectors and by the employment effect accompanying

the emergence of new sectors. We can then expect that as societies become richer, a growing

share of their income will be spent on (i) new products and services and on (ii) higher quality

and more differentiated products and services. As a consequence the distribution of product

quality and prices within the offered range changes with the average (or median) product

moving from the cheapest to higher priced ones. To test this hypothesis we carried out some

experiments in which we changed the extent to which product quality and differentiation are

increased by search activities. In the present version of TEVECON we cannot calculate demand

24

curves since the model, being designed to study long range phenomena, calculates one demand

point at each time. We can calculate quasi demand curves for extended periods of time of

variable length. We do this using different values of the parameters affecting product quality and

product differentiation. The results are shown in Fig 5, where we see that when search activities

lead to high rates of growth of product quality and product differentiation demand curves are

not uniformly downward sloping as price rises, but contain a section in which demand rises as

price rises. A similar behaviour is found for many types of differentiated goods, such as cars,

watches, television sets, washing machines, etc.

The section in which demand and price rise together correspond also to the behaviour observed

by Veblen (1899) (Trigg, 2001) and are called conspicuous consumption, according to which

consumers do not buy what they need in a functional sense but to impress other people and to

establish their social status. While before the industrial revolution this used to be true only for

very rich and powerful people, with the growing wealth generated in industrialized countries

such a form of consumption has become very common. Indeed, the cheapest goods in each

sector are not those which sell the most. Figure 7 shows that the possibility of conspicuous

consumption increases, when innovation generates high rates of growth of product quality and

product differentiation of goods and services. Given the extent to which this phenomenon has

become widespread, it is doubtful whether the term conspicuous consumption should be used

for the whole range in which demand rises with rising price or just for extremely expensive

products. Further experiments are required to clarify this type of behaviour.

- 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

D

P

25

Fig 7. Effect of changing the parameters affecting product quality (k14) and product

differentiation (k15) on the shape of demand curves. Product quality and product differentiation

increase in the order dark blue, light blue, violet, green, red. Demand on the vertical axis, time on

the horizontal axis.

5. SUMMARY AND CONCLUSIONS

In this paper we study the evolution of consumption activities by means of our TEVECON model

of economic development. In previous papers (Saviotti, Pyka, 2012, 2013) we showed that the

range, quality and internal differentiation of goods and services increased enormously since the

industrial revolution and in particular since the beginning of the 20th century. Furthermore, we

had already stressed that innovation alone cannot explain economic development unless

demand co-evolves with it. In this paper we analyze the mechanism by means of which demand

and innovation interact, and in particular the role of demand saturation in inducing the

emergence of new goods and services. We find that complete saturation, in which demand first

rises but then becomes constant as income rises, is a relatively infrequent behaviour. Deviations

from saturation, and consequently the shape of Engel curves, are affected by several factors. For

example, the presence or absence of an income term in the demand function leads to

considerable changes in the shape of Engel curves. Saturation of sectoral demand is never

achieved because the emergence of new sectors keeps shifting the potential saturation level

which is attained, if there was a constant number of sectors. Additionally, when the demand

function contains a disposable income term and there are two social classes differing for their

education levels and competencies, Engel curves show either over-saturation or under-

saturation, where the former leads to a fall in the rate of growth of demand with rising income

and the latter leads to a slowdown in the rate of growth of demand when income rises. We will

need more experiments to explore the potential impact of other variables and parameters on the

shape of Engel curves. However, even from these limited experiments, we can conclude that the

relationship between demand and income and the shape of Engel curves are not intrinsic

properties of demand, but that they depend on the interactions of demand with a number of

other variables of the socio-economic system. This reinforces our conviction that co-evolution is

a fundamental mechanism of economic development, which is essential to understand long run

processes of structural change and transformations.

REFERENCES

26

Aghion P., Howitt P. (1992), A model of growth through creative destruction, Econometrica, Vol. 60, 323-351.

Aversi R, Dosi G, Fagiolo G, Meacci M, Olivetti C (1999) Demand dynamics with socially evolving

Bils M., Klenow P.J.(2001), Quantifying Quality Growth, The American Economic Review, Vol. 91, No. 4, pp. 1006-1030

Boldrin M., Levine D.K., (2010), All the interesting questions, almost all the wrong reasons, in Pecchi L. Piga G. (Eds) Revisiting Keynes: Economic Possibilities for our Grandchildren, Cambridge MA, MIT Press

Bouvier G., Diallo F. (2014), Soixante ans de réduction du temps de travail dans le monde , INSEE http://www.insee.fr/fr/themes/document.asp?ref_id=ip1273

Chai, Andreas, Moneta Alessio (2010), "Retrospectives: Engel Curves," Journal of Economic Perspectives, American Economic Association, vol. 24(1), pages 225-40, Winter.

Chai, Andreas, Moneta Alessio (2011), "Back to Engel? Some evidence for the hierarchy of needs," Papers on Economics and Evolution 2011-13, Max Planck Institute of Economics, Evolutionary Economics Group.

Chai, Andreas, Moneta, Alessio (2012), Back to Engel? Some evidence for the hierarchy of needs. Journal of Evolutionary Economics (2012)

Engel E. (1857), Die Productions –und consumtions –Verhältnisse des Königreichs Sachsen , Bulletin de l’Institut International de Statistique9

Foellmi, R., Zweimuller, J., (2006), Income distribution and demand induced innovations, Review of Economic Studies 82, 95 112‐

Frenken K, van Oort FG,Verburg T (2007) Related variety, unrelated variety and regional economic growth, Regional Studies 41 (5): 685-697

Funke M, Ruhwedel R (2001a) Product variety and economic growth: Empirical evidence for the OECD countries, IMF Staff papers, 48, No. 2

Funke M, Ruhwedel R (2005) Export variety and economic growth in East European transition economies, The Economics of Transition 13(1): 25-50

Funke M, Ruhwedel R, (2001b) Export variety and export performance: empirical evidence from East Asia, Journal of Asian Economics 12: 493-505

Hidalgo CA, Klinger B, Barabasi A-L, Hausmann R (2007) The product space conditions the development of nations, Science 317(5837): 482-487

Imbs J., Wacziarg R., (2003), Stages of Diversification, American Economic Review, vol. 93, no. 1, p. 63-86.

Jevons WS (1924) The theory of political economy, 4th edn. MacMillan, London

Keynes J.M. (1928), Economic possibilities for our grandchildren, in Pecchi L., Piga G. (Eds), Revisiting Keynes: Economic possibilities for our grandchildren, Cambridge MIT Press (2008)

Lancaster K.J. (1966), A new approach to consumer theory, Journal of Political Economy, Vol. 14 (1966) 133-156.

Manig C., Moneta A. (2014), More or better? Measuring quality versus quantity in food consumption, J. Bioeconomics (2014) 16:155–178; DOI 10.1007/s10818-013-9174-4

Marshall A (1949), Principles of economics, 8th edn. MacMillan, New York

Matsuyama, K., (2002), The Rise of Mass Consumption Societies, Journal of Political Economy 110, 1035 70.‐

Menger C. (1950), Principles of economics, New York University Press, New York London

27

Nelson, R. Winter, S. (1982), An Evolutionary Theory of Economic Change, Cambridge, Mass, Harvard University Press, (1982).

Pasinetti L. L. (1981), Structural Change and Economic Growth. Cambridge, Cambridge University Press.

Pasinetti L.L., (1993), Structural Economic Dynamics, Cambridge, Cambridge University Press.

Pecchi L. Piga G. (Eds) (2008), Revisiting Keynes: Economic Possibilities for our Grandchildren, Cambridge MA, MIT Press

preferences. Industrial and Corporate Change 8: 353–399

Pyka, A., Saviotti, P.P. (2011), Economic Growth Through The Emergence of New Sectors, in Mann S. (Ed) Sectors Matter, Berlin- Heidelberg, Springer

Romer, P. (1990), Endogenous technical progress, Journal of Political Economy, Vol. 98, 71-102.

Saviotti P.P. (2001), Variety, growth and demand, Journal of Evolutionary economics, Vol.11 (2001) 119-142

Saviotti P.P., Metcalfe J.S. (1984) A Theoretical Approach to the Construction of Technological Output Indicators, Research Policy, 13 (1984), 141-151

Saviotti P.P., Pyka A. (2004), Economic development by the creation of new sectors, Journal of Evolutionary Economics, 14, (1) (2004) 1-35.

Saviotti P.P., Pyka A. (2008), Micro and macro dynamics: Industry life cycles,inter-sector coordination and aggregate growth, Journal of Evolutionary Economics, Vol 18 (2008) pp. 167-182

Saviotti P.P., Pyka A. (2012), On The Co-Evolution Of Innovation And Demand: Some Policy Implications, Revue de l'OFCE, n°124, pp 349-388 - Debates and policies: Agent-based models and economic policy, edited by Jean-Luc Gaffard and Mauro Napoletano

Saviotti P.P., Pyka A. (2013), From necessities to imaginary worlds: Structural change, product quality and economic development, Technological Forecasting & Social Change, Vol. 80 (2013) 1499–1512

Saviotti P.P., Variety, growth and demand, Journal of Evolutionary Economics (2001) 11: 119–142

Schumpeter, J. (1934, original edition 1912) The Theory of Economic Development, Cambridge, Mass, Harvard University Press.

Solow, R.M. (1956), A contribution to the theory of economic growth, Quarterly Journal of Economics, Solow, R.M., Vol 70 (1956) 65-94.

Solow, R.M. (1957), Technical change and the aggregate production function, Review of Economics and Statistics, Vol. 39 (1957) 312-320.

Stiglitz J.E. (2010), Toward a general theory of consumerism: Reflections on Keynes 's Economic possibilities for our grandchildren, in Pecchi L., Piga G. (Eds), Revisiting Keynes: Economic possibilities for our grandchildren, Cambridge MIT Press (2010)

Trigg, A. (2001), "Veblen, Bourdieu, and conspicuous consumption", Journal of Economic Issues, Vol. 35 No.1, pp. 99–115

Veblen, Thorstein B. (1899) The Theory of the Leisure Class: An Economic Study in the Evolution of Institutions (New York: Macmillan). Maslow, A. H. (1943). "A theory of human motivation". Psychological Review 50 (4): 370–396.

Walras L (1896) Elements d’economie politique Pure, 3rd edn. Economica, Paris (reprinted 1988)

Whaples, Robert (2001), “Hours of Work in U.S. History”. EH.Net Encyclopedia, edited by Robert Whaples. August 14, 2001. URL http://eh.net/encyclopedia/hours-of-work-in-u-s-history/

28

Witt U. (2001), "Learning to consume: a theory of wants and the growth of demand." Journal of Evolutionary Economics, Vol. 11.

Zilibotti F. (2010), Economic possibilities for our grandchildren 75 years after: a global perspective, in Pecchi L., Piga G. (Eds), Revisiting Keynes: Economic possibilities for our grandchildren, Cambridge MIT Press (2010).

29