The Rise of Mass Consumption Societies, Summaries of American literature

Download The Rise of Mass Consumption Societies and more Summaries American literature in PDF only on Docsity!

The Rise of Mass Consumption Societies

Kiminori Matsuyama

Northwestern University

Abstract

This paper studies mechanisms behind the rise of mass consumption societies. The development process depicted follows the Flying Geese pattern, in which a series of industries takes off one after another. As productivity improves in these industries, each consumer good becomes affordable to an increasingly large number of households, which constantly expand the range of goods they consume. This in turn generates larger markets for consumer goods, which leads to further productivity improvement. For such virtuous cycles of productivity gains and expanding markets to occur, income distribution should be neither too equal nor too unequal. With too much equality, the economy stagnates in a poverty trap. With too much inequality, the development stops prematurely.

Although Katona stressed that this is a phenomenon unique to the American society, virtually all the industrialized countries have gone through similar transformation after WWII.^1 Rostow (1960), in developing his thesis of stages of economic growth, named the last of the five stages, “the age of high mass consumption.” He argued that not only the United States, but also Canada, Australia, Western European countries, and Japan had reached this stage. Fourastié (1979) discussed similar development in postwar France, from 1946 to 1975, the period that many French writers call Les Trente Glorieuses after the title of his book. Many Japanese also commented on a new feature of postwar booms in the fifties and sixties; Contrary to the prewar booms, which were mostly driven by military demand, they were driven, or at least supported, by consumer demands, particularly for home electronic appliances.^2 One piece of the evidence that these authors routinely present is the penetration rates of consumer goods. Figure 1 illustrates the typical pattern in a stylized way. Each curve shows the fraction of households using a particular consumer good. For example, the use of vacuum cleaners, washing machines, telephones, was restricted to a small section of the population before WWII, but spread to the low-income households during the fifties and sixties. Many other consumer goods, such as television sets, cars, and air-conditioners follow similar paths, with some lags.^3 This pattern is so similar across many industrialized countries that the penetration rates of representative goods have become the popular yardstick for comparing the standards of living across societies. One key feature of this pattern is that not only the market for each consumer good takes off, but also each takeoff is followed by one after another. The pattern shown in Figure 1 will be called “Flying Geese” in this paper.^4

As many countries have experienced this transformation, the very notion of necessities and luxuries has changed. Many consumer goods that have penetrated into the majority of households, such as vacuum cleaners, washing machines, telephones, televisions, refrigerators, automobiles, air-conditioners, are now generally regarded as necessities in rich societies, and yet, they were all considered as luxuries only a half century ago. To quote Katona again,

“We are rich compared with our grandparents and compared with most other peoples of the world. In fact, however, we are still a middle-class society, enjoying middle-class comforts. …. The drudgery of seeking subsistence has been supplanted for millions of people, not by abundance and indulgence, but rather by a new concept of what are necessities and needs .” (italics added)

The notion of necessities and luxuries not only has changed over time. It also varies from countries to countries. Many goods that are taken for granted in rich countries remain luxuries in many parts of the world. The question of why some countries have failed to become mass consumption societies is at least as important as the question of why some succeeded. This paper develops a model, which is consistent with the key features of mass consumption societies described above, and then uses it to understand the mechanisms behind the rise of mass consumption societies, and to identify the conditions under which a country succeeds in making such transformation. What is central to the analysis is a two-way causality between productivity improvement and the rise of a mass consumption society. As productivity improves, the prices of consumer goods go down, and they become affordable to an increasingly

even though there is no interindustry spillover of learning-by-doing. The intuition behind this pattern is easy to grasp. The purchase of a good by the high-income households reduces its price, which makes this good affordable to the low-income households, which were previously unable to purchase it. This trickle-down process helps an industry to take off. However, this is not the end of the story. The purchase of a good by the low-income households, by pushing down its price even further, helps to reduce the expense of the high-income households. This allows them to purchase the next item on their shopping list. Through this trickle-up process, productivity gains in one industry lead to productivity gains in the next. Second, the set of steady states is a lattice, and the economy grows monotonically until it converges to the least element of the lattice. That is, if there are multiple steady states, the economy is trapped into the lowest steady state, where a relatively small fraction of the households consumes a relatively small set of the goods. Thus, there is the possibility that the trickle-down and trickle-up processes are interrupted. Third, the dynamic evolution of the economy depends critically on income distribution. Some income inequality is needed for the economy to take off; with too much equality, the economy stagnates in a poverty trap. This is because, in order to trigger the process, the economy needs a critical mass of the rich households, which can afford to buy some goods, even when they are still expensive. With too much inequality, on the other hand, the process stops prematurely. This is because neither trickle-down nor trickle-up mechanisms would work if there are too much income gaps. To put it another way, the rise of a mass consumption society requires income to be distributed in certain ways. Because of this, the effects of income transfer also turn out to be subtle. Perhaps the analogy of the dominos may be useful. In order for the

dominos to continue falling like a cascade, they need to be spaced appropriately. If they are put tightly together, the dominos cannot fall. If there is a big gap between dominos, a falling domino cannot knock down the next one, hence the chain reactions will be interrupted. It is worth emphasizing that the model developed in this paper explains the Flying Geese pattern based on endogenous technological changes. One might be tempted to argue that we observe the Flying Geese pattern simply because different consumer goods were invented at different times. Such an explanation based on exogenous technological progress has a couple of problems. First, while the penetration rates of consumer goods display similar patterns across many developed countries, their timings are different across countries. Indeed, these goods hardly have penetrated in many underdeveloped countries. Second, many consumer goods, such as vacuum cleaners, washing machines, telephones, radios, televisions, automobiles, were invented by the early twentieth century in their most primitive forms. Only through further improvement, these goods have become affordable to the majority of the households in developed countries. And the market size is one of the critical factors determining the speed of such improvement. This is not to deny the possibility that some major technological advances that were applicable to many industries, such as electric motors or Taylorism, were responsible for making the rise of mass consumption societies possible for the first time in human history. Any theory based on exogenous technological events, however, cannot explain why the United States led the way in becoming the mass consumption society nor why certain goods spread faster than others. There are a few related studies in the literature. Baland and Ray (1991) and Murphy, Shleifer, and Vishny (1989) both studied models of increasing returns and nonhomothetic

The rest of the paper is organized as follows. Section 2 sets up the model, and derives the dynamical system governing the evolution of the economy. Section 3 discusses some general properties of the system. Sections 4 and 5 look at special cases to examine in detail the roles of income distribution. Section 6 discusses alternative specifications of the model. Possible extensions are discussed in the concluding section.

2. The Model. This section describes the structure of the economy and derives the system that governs the dynamic trajectory of the economy. Many assumptions discussed below are adopted to simplify the exposition, and can be relaxed or replaced by alternative assumptions, as will be explained in Section 6. The economy is populated by a continuum of households with the unit measure_._ They supply labor, and consume some goods and leisure. Goods are produced by labor only. The detailed descriptions of the model are now given, first the preferences, then income distribution, and finally technologies.

2A. Goods and Preferences: There are J +1 goods, labeled j = 0,1,..., J. Good zero is food; it is a homogenous, divisible good. In addition, there are J manufacturing goods, indexed as j = 1,..., J. They are indivisible and come in discrete units. All the households have the same preferences, given by the following utility function:

c if c
1 U = 1
(^)  kJ
1  (^)
jk
1 x (^) j 

l if c > 1

where c is food consumption, l is the leisure, and xj is an indicator function, with xj = 1 if manufacturing good j is consumed and xj = 0 if it is not. Food is a necessity, and the household needs to consume a minimum amount of food, the subsistence level, before consuming any manufacturing good. The subsistence level is normalized to be one to ease the notational burden. It is also assumed, for the sake of simplicity, that the propensity to spend on food is equal to zero above the subsistence level. The preferences over manufacturing goods have the property that the households benefit nothing from consuming good k , if xj = 0 for some j < k. This implies that the households consume good k , only if they also consume all the manufacturing goods, whose indices are less than k. In other words, the households have a well-defined priority over the set of manufacturing goods, with a lower indexed good is higher on their shopping list. What is also implicit in the preferences is that the household’s demand for each manufacturing good satiates after one unit. It is worth emphasizing that neither the strong form of Pareto-Edgeworth complementarities nor the assumption that all the households have the same ordering are essential in the following analysis: see sections 6B and 6C for more detailed discussion. What is essential is that the households do not change their orderings, when the relative prices change. As long as the range of relative price movements is appropriately restricted, much of the results

I  PJ  c = 1, l = I  PJ , xj = 1 ( j = 1,..., J ),

where Pk = (^)
^ kj
0 pj can be interpreted as the minimum level of income that induces the

household to consume manufacturing good k. The most important feature of the individual demand curve derived above is that an additional income translates into an additional demand for a manufacturing good, only when it pushes the household’s income above the critical level of income. If the household’s income level is well below Pj , an additional income would be spent on food, leisure, or manufacturing goods with lower indices. For the poor, good j remains a luxury, which is beyond their means. If the household’s income level exceeds Pj , on the other hand, an additional income would be spent on leisure, or manufacturing goods with higher indices. For the rich, good j is a necessity, with which they are already satiated. What is essential for the following analysis is that the marginal propensity to spend on a manufacturing good is small when income is either very low or very high. This property of demand captures the following idea. A manufacturing good is a luxury for many at a lower level of economic development. As the economy develops and an overall level of income grows, it changes from a luxury to an amenity, and then to a necessity. In other words, the very notion of what is a luxury and what is a necessity changes with income.^7

2C. Income Distribution and Aggregate Demand Having derived individual demand curves, the next step is an aggregation. Let F be the distribution of income across households, i.e., F(I) is the fraction of the households, whose income is less than or equal to I. Income differs across the households due to skill differences,

reflected in differences in the effective labor supply. The total labor supply is thus equal to L =

0 IdF^ ( I ).^8 The income is the only source of heterogeneity across the households. Since only the

households, whose income is higher than Pj = (^)
(^) i^ j
0 pi purchase a manufacturing good j , and no

household purchases more than one unit of any manufacturing good, the aggregate demand for good j is equal to the mass of the households, whose income is higher than Pj :

(1) C (^) j = 1  F ( Pj ) = 1  F (^) ^
i
^ j 0 pi

. ( j = 1,..., J )

Many features of the aggregate demand functions, eq. (1), deserve emphasis. First, it depends on income distribution, because the marginal propensity to spend on a manufacturing good varies with the household income. Second, it is bounded from above by one. This is because the size of the market for a manufacturing good is limited by the number of households that can afford to consume it, not by the aggregate income of the economy. Third, a decline in the price of good i does not affect the demand for good j < i ( C (^) ji =  C (^) j /  pi = 0), while it generally increases the demand for good j  i ( C (^) ji  0). In other words, demand complementarity (in the sense of Hicks-Allen) exist from a lower indexed good to a higher indexed good, but not the other way around. This is because of the asymmetric way in which the income effect of price changes operates. A decline in the price of good i only affects the households, whose income is higher than Pi , and these households may respond by spending the increased real income on higher

where the dot indicates the time derivative. Note also that the depreciation keeps Qj ( t ) from growing unbounded. Indeed, as seen in (2), it is bounded from above by one, because C (^) j ( t ) is bounded from above by one. What is important here is the assumption that there exist some forms of dynamic increasing returns in each consumer goods industry. Learning-by-doing in production is adopted here because it is the simplest (and perhaps most standard) way of modeling dynamic increasing returns. It is also worth pointing out that there is an alternative interpretation of eqs. (2) and (3): learning-by-doing in consumption. The “price” of a consumer good that the household must pay includes not only the price charged by the producers, but also the effort required by the household to use the good. As more households accumulate experiences, the required amount of effort will decline, thereby reducing the effective price of the good, measured in leisure. Such dynamic consumption externalities would be isomorphic to learning-by-doing in production in the present model. The distinction between these two forms would be critical in an open economy (see Section 7).

2E. The Dynamical System. We are now ready to derive the dynamical system that describes the law of motion governing the trajectory of the economy. First, note that perfect competition in each industry ensures that the price of each good is equal to the marginal (and average) cost, which is nothing but the unit labor requirement. (Recall that labor is the numeraire). Therefore, we have

(4) p 0 = a 0 and pj ( t )= Aj ( Qj ( t )) ( j = 1,..., J ).

Inserting (4) into (1) yields

(5) C (^) j ( t ) = 1  F (^) ^ a 0 
i
^ j 1 Ai ( Qi ( t ))

 Dj ( Q ( t )) , ( j = 1,..., J )

where Q = ( Q 1 , Q 2 ,…, QJ )
[0,1] J. Let D ( Q )  ( D 1 ( Q ), D 2 ( Q ), …, DJ ( Q )). Note that Dij =

 Di /  Qj = 0 for all i < j and Dij  0 for all i  j due to the (asymmetric) demand

complementarity. Therefore, the mapping, D : [0,1] J [0,1] J , is increasing in that Q’ – Q 
J

implies D ( Q’ )– D ( Q ) 
J
, where
J
is the set of J -dimensional nonnegative vectors.

Inserting (5) into (3) yields

(6) Q^
(^) j ( ) t =  j { Dj ( Q ( t ))  Qj ( t )}  (^) j ( Q ( t )) , ( j = 1,..., J )

which can be expressed in a more compact manner, as follows:

(7) Q^
= ( Q )

where = ( 1 , 2 , …, (^) J ): [0,1] J 
J is a vector field on [0,1] J.

(P2): (^) ij   (^) i /  Qj = 0 if i < j ; (^) ii =  i (Dii  1); (^) ij =  i Dij  0 if i > j. The dynamical system is thus recursive and cooperative. It is recursive in that the dynamics of ( Q 1 , … , Qj ) is independent of that of ( Qj +1 , … , QJ ) for all j. The reason for this is asymmetry in which demand complementarity operates in this economy. As one industry improves its productivity and its cost and output price declines, only the industries with higher indices see demand for their goods increase. The resulting increase in output leads to a faster learning only in these industries. The system is also cooperative in the sense of Hirsch (1982), that is (^) ij  0 for all i j. The system is cooperative because productivity improvement and the resulting price reduction in one industry may increase but never reduces demand in other industries. It should be noted that the mechanism through which productivity improvement spillovers from an industry with a lower index to an industry with a high index is demand complementarity. In the present model, all the learning-by-doing effects are industry-specific. An industry learns nothing from manufacturing experiences of other industries.

Let

J
denote the set of J -dimensional vectors with positive components.

(P3): M + 
Q 
0 , 1  J^ ( Q )
J
=
Q 
0 , 1  J^ D ( Q ) Q 
J
, M ++ 
Q 
0 , 1  J^ ( Q )
J



=
Q 
0 , 1  J^ D ( Q ) Q 

J
, M – 
Q 
0 , 1  J^ ( Q )

J =
Q 
0 , 1  J^ Q  D ( Q )
J
, and

M –– 
Q 
0 , 1  J^ ( Q )
J

=
Q 
0 , 1  J^ Q  D ( Q )

J
are positively invariant.

That is, Q s^
( )
 J 



^ J^ implies Q t^
( )
 J 



^ J^ for all t  s , and
Q s^
( )  J

^ J^ implies^
^ Q t^
^ ( )^^  J



^ J^ for all^ t^ ^ s. This is another property of a cooperative system,

which maintains the monotonicity of trajectories. See Smith (1995, Proposition 3.2.1). Roughly speaking, it means that all the industries move together.

(P4): The set of steady states, S 
Q 
0 , 1  J^ ( Q )
0 =
Q 
0 , 1  J^ Q
D ( Q ), is a nonempty,

compact lattice, where the ordering is induced by

J. The greatest element of S is sup M + and its

least element is inf M –. This follows from applying Tarski’s (1955) fixed-point theorem to D : [0,1] J [0,1] J. A lattice is a partially ordered set, which contains both the least upper bound and the greatest lower bound of any pair of its elements. One important feature of a lattice is that, if it is compact, it contains both its greatest and least elements.

(P5): For any initial condition, Q (0) [0,1] J^ , lim (^) t
 Q ( t ) S. Thus, the system is globally convergent; the economy converges to a steady state without any exception. To understand this, note first that any one-dimensional dynamical system is globally convergent. Since the dynamics of Q 1 is independent of the rest of the system (the recursiveness), it can be viewed as a one-dimensional system, hence it converges globally. This effectively reduces the dimensionality of the system by one. Repeating this process shows that the global convergence of the entire dynamics. It is also worth pointing out that, even without recursiveness, a cooperative system is globally convergent if J
2 (Smith 1995, Theorem 3.2.2).