The Microeconomic Foundations of Business Cycles: From Institutions to Autocatalytic Networks

33
The Microeconomic Foundations of Business Cycles: From Institutions to Autocatalytic Networks Igor Matutinovic ´ In his seminal book Capitalism, Socialism, and Democracy, Joseph Schumpeter argued that the very nature of capitalism is economic change and that capitalism never was nor will it ever be stationary (1976). Large-scale fluctuations are one of the key aspects of nonstationary behavior of market economies where total output, employment, and investments exhibit aperiodic ups and downs of various durations and intensity. One of the unresolved questions in economic theory is whether business cycles are an inherent and thus irremovable feature of capitalist economies or they result from external shocks acting on economic systems otherwise poised at equilibrium. During unusually long phases of economic expansions, some distinguished mainstream economists have even proclaimed that business cycles might be gone forever (Fuhrer and Schuh 1998; Auyang 1998, 282; De Long 1999). During the last long expansion of the U.S. economy, an eco- nomic conference held in Boston in 1997 addressed especially this issue and most partic- ipants rejected the idea that the business cycle is dead (Fuhrer and Schuh 1998). However, there was no consensus on the causes of recessions past and future, except for the notion that an economy exhibits some kind of “vulnerability” which eventually trans- fers into a recession. This debated field of the origins of business cycles is stretched between exogenous, shock-based and endogenous, evolutionary-like theorizing. Theories and models that posit an equilibrium economic setting, disturbed by random technology shocks, like real business cycle theory, lack explanatory plausibility while they do not prove them- selves valid in predictions either (Auyang 1998; Dow 1998; Ormerod 1998). The pres- ent work, among other things, presents an implicit critique of such theorizing. 867 JOURNAL OF ECONOMIC ISSUES Vol. XXXIX No. 4 December 2005 The author is with the GfK Center for Market Research in Zagreb, Croatia. He is grateful to João Filipe, Dias Rodrigues, Tommaso Luzzati, and Velimir Pravdic ´ for reading and commenting on the earlier version of this text. He thanks Robert Ulanowicz for commenting on the section on autocatalysis and two anonymous reviewers for their valuable suggestions. He also thanks the GfK Group for the support of this research. Jei © 2005, Journal of Economic Issues

Transcript of The Microeconomic Foundations of Business Cycles: From Institutions to Autocatalytic Networks

The Microeconomic Foundations of Business Cycles:

From Institutions to Autocatalytic Networks

Igor Matutinovic

In his seminal book Capitalism, Socialism, and Democracy, Joseph Schumpeter argued that

the very nature of capitalism is economic change and that capitalism never was nor will it

ever be stationary (1976). Large-scale fluctuations are one of the key aspects of

nonstationary behavior of market economies where total output, employment, and

investments exhibit aperiodic ups and downs of various durations and intensity. One of

the unresolved questions in economic theory is whether business cycles are an inherent

and thus irremovable feature of capitalist economies or they result from external shocks

acting on economic systems otherwise poised at equilibrium. During unusually long

phases of economic expansions, some distinguished mainstream economists have even

proclaimed that business cycles might be gone forever (Fuhrer and Schuh 1998; Auyang

1998, 282; De Long 1999). During the last long expansion of the U.S. economy, an eco-

nomic conference held in Boston in 1997 addressed especially this issue and most partic-

ipants rejected the idea that the business cycle is dead (Fuhrer and Schuh 1998).

However, there was no consensus on the causes of recessions past and future, except for

the notion that an economy exhibits some kind of “vulnerability” which eventually trans-

fers into a recession.

This debated field of the origins of business cycles is stretched between exogenous,

shock-based and endogenous, evolutionary-like theorizing. Theories and models that

posit an equilibrium economic setting, disturbed by random technology shocks, like

real business cycle theory, lack explanatory plausibility while they do not prove them-

selves valid in predictions either (Auyang 1998; Dow 1998; Ormerod 1998). The pres-

ent work, among other things, presents an implicit critique of such theorizing.

867

JOURNAL OF ECONOMIC ISSUES

Vol. XXXIX No. 4 December 2005

The author is with the GfK Center for Market Research in Zagreb, Croatia. He is grateful to João Filipe, Dias Rodrigues,

Tommaso Luzzati, and Velimir Pravdic for reading and commenting on the earlier version of this text. He thanks Robert

Ulanowicz for commenting on the section on autocatalysis and two anonymous reviewers for their valuable suggestions. He

also thanks the GfK Group for the support of this research.

Jei

© 2005, Journal of Economic Issues

Endogenous theorizing, which originated with classical economists, maintains its

tradition in institutional (evolutionary) and (post) Keynesian economics. Endogenous

theorizing often has been directed to the real causes of business cycles—those related to

production and consumption. This line of thought started with Karl Marx and probably

had its creative peak with Wesley Mitchell, John Maynard Keynes, and Joseph

Schumpeter.1 One branch of these theories—the trade cycle theories, or theories of real

macroeconomic disequilibrium—which expanded the analysis as laid down by Marx and

others rather than on neoclassical views, addressed especially the overcapacity problem

(Screpanti and Zamagni 1993, 219). The common theme of these theories can be sum-

marized as follows: in the upturn phase, the excess of savings, opening of new markets,

or technological innovations stimulate expansion of investments and consequently the

growth of employment, prices, and profits. Driven by rising prices and profits, invest-

ments continue to expand until the production of capital and consumer goods exceeds

its demand. At this point, profits decrease, bank loans contract while interest rates go

up, and investments come to a halt, eventually pushing the economy into recession

(219–247). Creation of productive overcapacity is viewed as the main trigger of reces-

sions in trade cycle theories,2 but precisely how this overcapacity arises from agents’

interactions is not explained. This is, in my view, the general shortcoming of most

endogenous theories—they try to explain the origins of business cycles by relying exclu-

sively on aggregate variables like demand, money and credit supply, savings and invest-

ments, prices and wages, and so on but fail to provide a plausible micro-mechanism that

produces, again and again, the very fluctuations in these and other relevant macro-vari-

ables. Moreover, the explanatory and explained variables mostly coincide, and this

surely does not contribute to theoretical clarity.

Trade cycle theories, and likewise later neoclassical theories, did not include the

institutional dimension in their explanation of business cycles. If we consider an eco-

nomic system as a “coordinated set of formal and informal institutions” (Dallago 2002)

which direct and constrain agents’ behavior, then the failure to account for these may

constitute a serious methodological shortcoming. I refer hereinafter to Malcolm

Rutherford’s definition of institution as “a regularity of behavior or a rule that is gener-

ally accepted by members of a social group, that specifies behavior in specific situations,

and that is either self-policed or policed by external authority” (1996, 182). Mitchell

argued that business cycles are “unintended and perverse consequence of behavior pat-

terns created by society’s monetary, business and other institutions” (1927, in

Rutherford 1996, 41). He also outlined profit making as a central process influencing

all other activities in a business economy ([1927] 1954). The imperative of profit making

for private enterprises that are run by professional managers is in fact one of the central

institutions of a capitalist economy. Besides being a generally accepted rule of efficiency

of business activities and, therefore, a major criteria of selection among different

options with different social, environmental, and economic outcomes, it also specifies

behavior in specific situations such as, for example, cost-cutting measures and reduction

of output in times of falling profitability. Cutting prices to the “market-clearing level” of

868 Igor Matutinovic

stock and productive capacity is theoretically possible, but it does not guarantee that

profitability will eventually be reached as compared with the cost cutting, which is under

the full control of a firm. Management adherence to such a behavior is usually policed

by supervisory boards and institutional investors (financial markets), which react posi-

tively to official announcements of cost-cutting plans such as large lay-offs and plant clo-

sures. Another important issue to be considered here is the managerial pursuit of

business growth, which is an institution insofar as it represents a generally accepted rule

of behavior in the business community. John K. Galbraith asserted that “the primary

affirmative purpose of the technostructure is the growth of the firm” as larger firms with

substantial market share gain better control over their costs, prices, suppliers, and

consumers (1973, 96). Besides that, growth serves the pecuniary interests of managers

while the large size of the firm shields them against takeovers from competitors.

The next relevant issue in our discussion of business cycles is the feedback mecha-

nism, which propagates change among economic agents and provides dynamic impetus

to prosperity and recession phases at the economy level. Mitchell identified a “cumula-

tive process” in the diffusion of business activity:

And each enterprise that finds its own trade increasing becomes an agency in

extending activity to still other enterprises. . . . Thus the revival gathers momen-

tum as the industries that receive a mild stimulus one after the other begin to

react upon those in which the movement started. (1941)

This cumulative process works also in reverse and has its modern conceptual coun-

terpart in autocatalysis, which will be discussed in detail later. The intensity of change

propagation in an economy is dependent on the size of firms which originate or trans-

mit a positive or a negative change in the business activity. Mitchell briefly pointed to

the importance of large-scale enterprise and specific industries, which he called centers,

in the causation of cycles ([1927] 1954)—an issue which will be addressed in the section

dealing with economic networks, “hubs,” and degree distribution of size of firms.

Finally, there is the issue of expectations and uncertainty and their effects upon the

cycle. Mitchell pointed out that “increase in profits during the expansion phase, com-

bined with the prevalence of business optimism, leads to a marked expansion of invest-

ments” (1941, chap. 5). Later he stated that “the twist given by overconfidence to

forecasts of future demand, always difficult to make with accuracy, thus leads in every

period of prosperity to an overstocking of certain markets.” Overheated optimism,

which results in illusory expectations about future returns of investments, is one of the

central themes in Keynes’ discussion of business cycles (1973a,b). Illusory expectations

about the future behavior of markets go hand in hand with the lack of forecast accuracy

of economic agents, which brings about two basic types of uncertainty that constrain the

rationality of their business decisions. The first is known as “operative indeterminacy”

(Monod 1970) or “procedural uncertainty” (Dosi and Egidi 1991, in Dequech 2001),

and it basically reflects the gap between theoretically available information and informa-

tion reasonably obtainable for a decision maker with respect to the time and cost of its

Microeconomic Foundations of Business Cycles 869

retrieval from the environment. The second type of uncertainty, usually referred to as

fundamental (also “essential” and “substantive” (Monod 1970; Dequech 2001), is char-

acteristic of Keynesian tradition and has a profound impact on long-term expectations

for the purposes of investment decisions (Fontana and Gerard 2004). It refers to matters

for which “there is no scientific basis on which to form any calculable probability what-

ever” (Keynes 1973a, 113–114, in Fontana and Gerard 2004), including long-term

changes in politics, technology, and consumer tastes. Both types of uncertainty prevent

economic agents from maximizing behavior while the latter introduces systemic

ambiguity in the evolution of single markets and in the economy as a whole.

In this paper I will try to show how a systemic interplay between two main

forces—institutional arrangements and autocatalysis, under the constraints of uncer-

tainty and nonconvex production costs, suffice to produce intermittent waves of pros-

perity and recession. I also hope to demonstrate that the bottom-up approach, which

explains large-scale fluctuations solely from institutionally constrained agents’ interac-

tions, is a promising methodological direction and that the role of overcapacity in busi-

ness cycle theorizing has not been fully exploited yet. The second aim of the paper is to

provide a microeconomic explanation as to why recessions must occur in a capitalist

economy. Actually, I will claim that recessions are inherent in the workings of market

economies.

The paper is organized as follows: the first section briefly describes autocatalysis and

small-world networks; the second section proposes the overcapacity theory of business

cycles; the third section presents empirical results; and the fourth section ends with dis-

cussion and conclusions.

Autocatalysis and Small World Networks

Assuming that economists are not necessarily familiar with small-world networks

and autocatalysis, what follows is a very short introduction to these concepts.

Small-World Networks

Complex systems can be represented as networks in which vertices or nodes denote

system elements, and directed edges or links the interactions or flows between them.

The orderliness of networks covers the continuum from ordered to random, whereas

complex systems are usually identified in between these two extremes (Watts and

Strogatz 1998). Among different statistical properties that usually define the degree of

orderliness or structure of networks (Watts and Strogatz 1998; Dorogovtsev and

Mendes 2001; Girvan and Newman 2001), two are of particular interest for this work:

1. Degree distribution. The fraction of nodes P(k) that have k or more links (both

incoming and outgoing). In many large networks, degree distribution has been

870 Igor Matutinovic

found to follow power law P(k) ~ k�� , where the number of nodes scales

inversely with their connectivity according to exponent � (Barabási and Albert

1999) (see figure 1).

2. Community structure. Subset of vertices within which vertex-vertex connections

are dense but between which connections are less dense. In social networks,

communities might represent groupings by common interest or a professional

background. Communities may exhibit a nested hierarchical structure

(communities within communities) (Girvan and Newman 2001).

Yet another measure of network organization used in process ecology (Ulanowicz

1997)—the interactions strength distribution—refers to the fraction of links P(ks) of a node

with intensity of flow {s � S} averaged over the whole population of nodes. It provides a

statistical measure of the weight of links incoming or outgoing of any node.

A particular network topology, called “small-world”3 network, has been proposed

by Duncan Watts and Steven Strogatz as a generic model for large, interactive natural

systems (1998). The small-world network topology is characterized by two basic proper-

ties (Watts and Strogatz 1998): (1) its characteristic path length L is short compared with

the size of the network (the “small world” effect) and is close to that found in random

networks4 Lsw �Lrandom and (2) its clustering coefficient, which shows the probability that

two nearest neighbors of a vertex are nearest neighbors also of one another, is much

higher compared with the random network Csw >> Crandom. Small-world networks dis-

play enhanced signal-propagation speed, computational power, and synchronizability

(Watts and Strogatz 1998). There are three structural classes of small-world networks

with respect to degree distribution (Amaral et al. 2000): (a) scale-free networks, (b)

broad-scale or truncated scale-free networks, where linear regime P(k) ~ k–� changes to

Microeconomic Foundations of Business Cycles 871N

um

be

r o

f n

od

es

N u m b e r o f lin k s p e r n o d e L o g s c a le

Lo

g s

cale

Figure 1. Power Law Distribution of Node Linkages

Note: The figure on the left represents characteristic size-frequency distribution of node linkages in many complex networks,

which is known as Pareto or power-law. The figure on the right is its representation on the logarithmic scale. As most nodes

have only a few links and some nodes have a large number of connections, the system has no characteristic scale. For an ac-

cessible overview of scale-free networks see Barabási and Bonabeu 2003.

exponential after a threshold connectivity k > kn has been surpassed, and (c) single-scale

networks, characterized by a connectivity distribution with a fast-decaying tail, such as

exponential or Gaussian. Small world effect and power-law degree distribution has been

found in a wide range of complex systems, like food webs in ecologies, protein and genes

in genetic networks, “html” documents on the World Wide Web, electric power grids,

and acquaintances in a social networks (Watts and Strogatz 1998; Barabási and Albert

1999; Montoya and Solé 2000; Girvan and Newman 2001; Dorogovtsev and Mendes

2001; Barabási 2002; Barabási and Bonabeu 2003).

Out of several known ways to model the evolution of networks which lie in the zone

between regularity and randomness (Dorogovtsev and Mendes 2001), I will mention

here only the process of preferential attachment (Barabási and Albert 1999), in which new

vertices attach preferentially (with higher probability) to the more highly connected ver-

tices in the network. Unlike other network models that use rewiring, this process func-

tions as an open system in which new vertices can be added continuously to the evolving

network. Preferential attachment operating in an open system resembles the positive

feedback process that operates in certain competitive market situations (see, for exam-

ple, Arthur 1990). Preferential attachment results in networks with power-law degree

distribution at the tail ends of which we find “hubs,” or nodes, with anomalously high

numbers of links, which in real networks have the function of keeping them from falling

apart (Barabási and Albert 1999; Barabási 2002). These networks are robust to an acci-

dental node failure but are highly vulnerable to a coordinated attack against their hubs

(Barabási and Bonabeu 2003). Richard Solé et al. (2003) have conjectured that the ori-

gin of power-law structure in disparate complex networks lies in the optimization

process, where reliable communication at low cost shapes network architecture.

Autocatalysis

Autocatalytic processes are ubiquitous in nature and lie at the origin of organiza-

tion in ecosystems at least as far as energy and material flows are considered (Ulanowicz

1997). In economics their presence and relevance has been more recognized since W.

Brian Arthur’s (1990) work on positive feedback and lock-in in competing technologies.

Some recent works, drawing from the analogy with chemistry, model economic growth

and development as autocatalytic processes emerging among firms that exchange skills

and products (Padgett et al. 2003). Here I will borrow from “process ecology”

(Ulanowicz 1997) and briefly describe certain properties of autocatalytic processes that

are also relevant for economic dynamics. Autocatalysis refers to “any cyclical concatena-

tion of processes wherein each member has the propensity to accelerate the activity of

the succeeding link” (Ulanowicz 1999, 41–55). Therefore it engenders and stimulates

growth of its constituent members until some physical or spatial constraint has been

reached and a negative feedback stops the process of growth, allowing the system to set-

tle in a meta-stable state. Autocatalysis stimulates competition and selection in the sense

that it preferentially streamlines energy and material flows toward more efficient mem-

872 Igor Matutinovic

bers, be these already inside the loop or acting at its periphery as potential “new

entrants.” In the competitive process a new member who contributes more to the

growth of the loop may replace a less efficient member. In that sense autocatalysis is

autonomous of its microscopic constitution: although individual members can be

replaced, the loop itself may persist over time. Autocatalysis imposes organization on a

system, which can be recognized, among other things, in the asymmetric distribution of

flows among its members. Because of their inner dynamics and openness to the wider

system, autocatalytic assemblages exhibit centripetality in amassing of material and

energy from their environment. Finally, as a consequence of competition and selection,

autocatalysis tends to ratchet all participants toward higher levels of performance

(Ulanowicz 1997, 46). This process is not confined to a single loop but transfers its influ-

ence to the wider systemic environment via connections that exist among assemblages of

different autocatalytic loops. One example of autocatalytic process in an economy,

which comprises all of the above-mentioned properties, is the formation of industrial

clusters (Porter 1990).

Autocatalytic loops commonly exhibit lags in time, and the time it takes one state

variable to respond to changes in another may cause overshoot followed by undershoot

(Cocks 1999). Cocks uses the business cycle as an example of such an overshoot-under-

shoot dynamics that arises as a consequence of the lag between the change in consumer

demand and production of goods necessary to meet this demand. As we shall see in the

next section, autocatalytic processes may be essential for the emergence of both order

and fluctuations in economic dynamics.

Overcapacity Theory Revisited

Network Structure of an Economy

An economy is a complex system consisting of a myriad of agents that may be placed in

three broad categories: firms, households, and government. Agents’ interactions come

under the broad umbrella of cooperation and competition while their production and con-

sumption activities constitute the functional fabric of the economic system. Economic activ-

ities often span several hierarchical levels of functional interdependence. For example,

production of cars includes firms from the primary sector (extractive and energy industry),

intermediate sector (processed materials, parts, tools, and machinery manufacturing), final

goods (vehicle producers), and a variety of services, including banking and distribution.

Functional connections among economic agents can be represented as a directed network

in which each vertex represents an agent and edges refer to functional relationships. In

essence, these functional relationships boil down to trading connections, where incoming

links ka refer to supply flows and outgoing links kb to selling flows, so that a degree k of a ver-

tex is the total number of its connections, k = ka + kb. A network model of the economy

assumes vertices of three kinds: (1) firms, which are connected to their suppliers, buyers, and

Microeconomic Foundations of Business Cycles 873

category markets; (2) category markets of distinct consumer product and service categories

(such as dairy products, passenger cars, OTC pharmaceuticals, apparel, and consumer elec-

tronics) that subsume household demand and represent households as an agent type in the

network; and (3) government as the single largest hub with many supply links incoming

from firms that spread across the entire network, representing its role of important buyer for

many sectors of economy. I assume that the economy network has the following four

properties:

1. It has a small-world effect, meaning that the average path connecting any two

agents in the economy is short, L � Lrandom.

2. The number of trading connections is not distributed uniformly among firms.

The probability that a firm i (i = 1,2,3,…n) has the number of trading

connections k, k � K, is power-law distributed:5

P(ki �K) ~ ki–�. (1)

This means that the number of firms in an economy that have a large number of

incoming links (suppliers) or outgoing links (buyers) will be relatively small.

These firms, which Albert Barabási (2002) termed “hubs,” are in fact analogous

to keystone species in ecosystems and they play a special role in the stability of

an economic network. They can be found among top Fortune 500

manufacturing firms, and a myriad of medium and small businesses depend on

their operation. So far the empirical evidence of scale-free degree distribution

in an economy has been found in the pharmaceutical and biotech industry in

Italy (Barabási 2002, 207) and the financial sector in Japan (Souma et al. 2003),

but there are reasons to believe that this connectivity pattern is ubiquitous in all

industrialized economies. We can infer this from another systemic property of

economies—size distribution of firms. It has been long known that size

distribution of firms is highly skewed (Simon and Bonini 1958), but conclusive

empirical work has been completed only recently. Using the criteria of output

and number of employees, Robert Axtell (2001) has demonstrated that the size

of firms in the U.S. economy is power-law distributed. As other empirical works

suggest (Ramsden and Kiss-Haypál 2000), this is most probably a common

pattern in all industrialized market economies.6 It is reasonable to suppose that

large firms have more business links than small firms in the same way that they

have more employees and output, so the degree distribution ki–� may be just an

aspect of the power-law size distribution of firms.

3. The intensity distribution of economic flows among firms in economy is highly

skewed. It can be defined as the interaction strength of business links—the share

of overall business of a firm that is conducted with each of its partners, both in

incoming and outgoing flow directions. I assume that the interaction strength

874 Igor Matutinovic

of business links in any firm is so distributed that the number of its partners

with the share of business S �s decays as power-law, N(Si) �Si–�, where Si denotes

a weighted business link of a firm i. Although such business data are not

publicly available, an indication that this statistical pattern may be true is given

by the well-known 80/20 rule of thumb, meaning that 80 percent of the

business of a firm is usually being conducted with only 20 percent of its

customers. According to Barabási (2002, 72) the 80/20 principle goes back to

Pareto distribution of wealth and has been related indeed to power-law

distributions. The generalization to the entire economy is straightforward. If we

look at the interaction strength of a single business link in its absolute terms, say

the dollar value of trade flowing through it annually, and if we comprise all

links existing in the economy by summing across all firms,i

n

iS�� 1 , we obtain

the probability distribution of value-weighted business links in the economy:

P S s S* ( )� � �� (2)

where � represents the average value of all individual exponents �. This means

that most business links in an economy network support trade flows of minor

value and that the number of links with high value throughput is comparatively

small. While individual links may change their strength over time, the

power-law distribution of a link’s strength remains a stable property of economy

network.

4. An economy is an ensemble of communities where each market represents a

distinct community in terms of a particular, dense topology of connections that

reflect functional (supply chain) interdependence among manufacturers and

include respective category markets if there exists a final consumer product.

Take, for example, the passenger car market: it is a community of densely

connected vertices representing specialized suppliers (materials, parts, and

tools), car manufacturers, car dealers, and a passenger car category market (one

node symbolizing household demand for cars). Sparsely distributed

vertex-to-vertex connections denote intermarket links, for example, between car

manufacturers and the chemical industry (dyes) or between car dealers and the

financial service industry (figure 2). A single, firm-to-firm or firm-to-category

market connection within a community denotes a niche, which is a narrowly

focused, specialized part of a market. Community structure is a looser way of

specifying a connectivity pattern between agents than clustering, and I see it as

more appropriate for modeling interactions in economy. Clustering would

imply improbable connections, like trading links among competitors7 and

between intermediate producers and category markets. The functional and

spatial self-organization of firms known as industrial clusters (Porter 1990)

shares some analogy with network clustering, but clustering topology cannot be

Microeconomic Foundations of Business Cycles 875

applied to the whole economy because of the above-mentioned incongruities.

Therefore, an economy is not strictly a small-world network as defined by Watts

and Strogatz (1998).

To wrap up the previously mentioned properties of an economy network: first, at

the economy level firms with small numbers of links (power-law degree distribution)

abound and among all existing business links those with low degree of strength or

low-value trading throughput (power-law interaction strength distribution) are perva-

sive. This is just another way of expressing the empirical fact that firm sizes are

876 Igor Matutinovic

Figure 2. Economy Network

Notes: Empty circles denote firms, the black circle is the government sector, and gray circles are category markets that represent

household demand. Firms are grouped in communities each representing a distinct market, where strong links symbolize

their functional interdependence. Weak links (dashed links) represent trade flows with other markets that do not represent

their core business. Each community of firms operating in a given market is connected at least with two links to its category

market, symbolizing competition of consumer goods manufacturers for household demand. Government is connected with

all three markets, but it has a strong link with only one of them. There are no functional connections among distinct category

markets and between them and the government hub.

power-law distributed. Second, if we look at the tails of these power-law distributions

and map the firms with connectivity k >> K and interaction strength s >> S with their

respective industries, we will identify hub industries or the market communities which

are more densely connected to other markets and where a fraction of these links are

strong ones. For example, the automotive industry and the construction industry (hous-

ing and plant building) buy from many different industries and sell to virtually all other

market communities. These hub industries play an important role in generating and

spreading disturbances across the economy network. Note here that at whichever scale

we choose to observe—firm, market, or economy—we find the same pattern of high skew-

ness in the observed variable, which is a general characteristic of far-from-equilibrium

self-organized systems.

Due to the small-world effect and community structure as described above, an econ-

omy is integrated into a tight whole. Such a network pattern has nontrivial implications

on an agent’s interactions and propagation of disturbances throughout different mar-

kets and up to the entire economy. Among other things, it confers nonlinearity to eco-

nomic flows between firms: a change in output of firm A will have considerable effect on

a small fraction of its business partners and a decreasingly smaller influence on the

remaining majority. It also implies that any disturbance, such as decrease in output in a

subset of firms, will eventually spread throughout the entire economy. The intensity and

the speed of the disturbance will be comparatively higher when a significant change in

output happens in a hub industry. The major impact of a single disturbance travels

along strong links that are pertinent to keystone or hub firms, and it spreads across mar-

kets in an ever-wider cascade of links. Suppose that a large consumer goods manufac-

turer cuts production by �q as a response to a drop of demand in its category market.

This will result in an adequate reduction of orders along its supply chain. A tiny fraction

of a manufacturer’s supply links is connected to large and medium-size companies (these

are strong links), and the reduction of orders will cascade downward, eventually reach-

ing small-size firms. In this cascade only a few firms will be affected significantly, but a

large number of firms will notice a change in their orders and output. Therefore, in such

a network, and to some extent analogously to Leontief’s input-output matrix, we can fol-

low how a change in output �Qij /�t of the i-th firm in the j-th market affects the output

of all other firms, spreading from a single market to the whole economy. Note that if the

network were a random one and the links were unweighted, then a disturbance starting

at any node would spread evenly throughout the economy, creating an extremely unsta-

ble and erratic environment. Conversely, on the ordered network there would not be a

sufficient flow of information necessary for coevolution and the economy would have

insular properties.

Agents’ Behavior

Agents are constrained in their problem-solving process by institutions that a priori

favor certain choices over other. Concerning the behavior of firms, I outline the impera-

Microeconomic Foundations of Business Cycles 877

tive of profit making, preference for growth of technostructure, and cost cutting as a

habitual measure to restore profits in times of business difficulty.

Firms are constantly involved in a decision-making process concerning choices of

production, investment, and marketing strategy. I assume that it follows the form of the

triadic problem-solving process, which is common to living beings in general: P1� TT1

�EE

P2 � TT2 (Popper 1994). Problems, P1, are confronted with tentative theories

(solutions), TT1, that undergo a selection process or error elimination, EE, eventually

leading to acceptable solutions and again to new problems, P2, that are inherent in the

solutions themselves. The choice of such a decision-making process implies that firms

operate in an environment of procedural and substantive uncertainty and that their

behavior is dominated by trial and error procedures rather then by other types of ratio-

nality. Triadic problem solving implies continuous generation of new goals and new

skills (technological or marketing tentative solutions to problems) that at any time may

result in a new behavior, at least in the subset of firm’s population. Karl Popper’s prob-

lem-solving process in itself provides a momentum for change and novelty in the econ-

omy: new aims spontaneously arise, problems solved with tentative solutions create

new, unintended, or unexpected consequences (i.e., new problems), and the process

repeats itself to infinity. By selecting such a decision-making procedure we can bridge

the gap between the two hierarchical levels—that of the individual and that of the

firm—and thus provide a crude psychological background of firm’s behavior. It follows,

then, that all fluctuations and novelty which arise at higher hierarchical levels—those of

single markets or economies—originate from a dynamic problem-solving process among

the principal market actors—firms and households. Therefore, exogenous shocks or any

other events outside an agent’s interactions are not considered here and are in fact

superfluous. Although I do not deny that exogenous shocks in principle may exist,8

their explanatory power concerning the long-term system behavior is irrelevant.

Evolution of Markets

Logistic growth has been found in a variety of social and economic phenomena

(Casti 1994), among others in the spatial and temporal diffusion of innovations, tech-

nologies, and transport infrastructure (Marchetti 1983; Grübler 1996). Evolution of

markets as a whole or a life cycle of individual products/services in any given market fol-

lows empirically well-established stages of emergence, growth, maturity, and decline or

rejuvenation, whose trajectory is usually described with a logistic curve (Kotler 1997,

364–367).

Suppose that a potential new market has been identified (e.g., personal computing)

and that a firm i introduces a product that serves the widest consumer base. As the new

product slowly gains ground, other firms, attracted by the novel business opportunity,

enter the market and the competitive struggle begins. In the competitive process, mar-

keters intensively research consumers in order to learn about their preferences, which

878 Igor Matutinovic

they afterward incorporate into tangible and intangible features of their products or ser-

vices. On the other side they strive to influence consumer preferences through advertis-

ing campaigns, whose main scope is to create actionable bias toward their particular

brands. Literally, producers adapt to and at the same time manipulate the structure and

the dynamics of demand by seeking to establish a positive feedback loop between them

and their clients: the more their products or services are present in the market now (mar-

ket penetration), the more likely it is that their sales will increase in the future, and this

constitutes the process of autocatalysis. The motion of demand for the products of firm i

can be approximated with a modified logistic difference equation:

D t r Q t CQ t

a tbQ ti j i i

i

i

i� � ��

��

��� �1

( )

( )( ) (3)

where Q refers to output, r is the average growth rate of the j-th market, Ci is the maxi-

mum capacity of i, ai is the individual coefficient of autocatalysis for the firm i, and b

refers to sales lost to competitors expressed as a fraction of its output. Note that demand

and output in the period t + 1 are practically equated and that the term Di can be substi-

tuted for Qi(t + 1) without changing the meaning of the equation. This is logically consis-

tent until productive capacity Ci reaches its theoretic maximum.

Autocatalytic term ai represents the average strength of entailment between the

firm and its clients, which determines its individual rate of demand at any discrete time

t: a tN

U ti i

l l

l

N

( ) ( )� ��

�11

1

� where U i

l denotes perceived economic contribution (utility,

benefit) of the firm’s i output to its l clients, and �l is a coefficient of proportionality,

denoting the relative volume of business done with the l-th client. In the following analy-

sis I assume, for the sake of simplicity, that ai(t) = const. or fixed throughout the market

evolution stages and that its value is defined on the interval (5,10) where 5 denotes the

minimum autocatalytic entailment and 10 its maximum. The parameter b governs the

effect of competition on the demand for products of the firm i. The case b > 0 reflects a

common competitive situation where a portion of demand goes to competitors. Com-

petitors constrain the firm’s capability to use its capacity efficiently: for all cases where b

> 0.15, firm i uses its capacity below minimum efficiency and is forced to close down.

The special case b = 0 means that firm i is a monopolist in the classical sense. The

monopoly may also be temporary as in the case of the market introduction phase when

an entirely new product is offered and competitors have not yet entered, or it can be nar-

rowly located in a market niche where a firm has no direct competitors because the mar-

ket is too small.

Equation (3) captures some basic market dynamics and situations.9 The average

market growth rate, r, governs the overall dynamics of demand for the product of firm i.

Under very low growth rates (r < 1.15) the market fails to kick off past its emergence

stage. Firms are not able to reach minimum capacity utilization rates, and the market

Microeconomic Foundations of Business Cycles 879

eventually dies off. This may be the case of many niche markets. When initially r > 1.15

but changes to r � 1 after demand for product category j has reached the saturation

point at �Dj/�t = 0 in the mature stage, it means that generic demand for a product cate-

gory is actually below its replacement rate (due to the product’s life cycle end and dis-

posal) and the market fades out (logarithmically) in the decline stage. Next, consider the

meaning of variations of the autocatalytic term under the condition (b = 0.1) that

denotes a strong competitive environment: if a = 5, demand for the firm’s products is

too low to bring its capacity utilization to the required efficiency minimum (I assume it

to be no less than 65 percent). This means that under sharp competitive circumstances

firms must strive for much higher autocatalytic entailment (or value adding) than the

minimal one. Even with its best efforts (a = 10), a firm cannot surpass the capacity utili-

zation threshold of 80 percent, which is probably close to the average value in most

industries. However, if b = 0 (the monopoly situation), the required minimum (a = 5) is

enough to reach a satisfactory level of 80 percent capacity utilization. When the compet-

itive environment is more relaxed (b = 0.05), firms can gain much by improving their

contribution to the value chain and at a = 9 they can reach theoretic maximum of capac-

ity utilization. Finally, if a = 0, then demand is zero, too, meaning that a firm was a fail-

ure from the very outset. In any case, firms that do not manage to establish strong

autocatalytic loops with their clients disappear quickly—they are being expelled by other

firms that contribute more value to the loop.10 The autocatalytic term can be related to

the problem of differential growth rates among firms. It was empirically shown for the

U.S. economy that growth rates of firms are power-law distributed, dependent on the

size of firm, and correlated in time (Stanley et al. 1996)—all contrary to the predictions

of Gibrat’s law. I conjecture that these differential growth rates among firms arise from

the different intensity of autocatalytic entailment that firms establish with their clients

during the market growth stage. By the time a market approaches its mature phase, these

differential growth rates have produced characteristic asymmetric distribution of mar-

ket shares.

Let us consider now what happens with the evolution of the total market. Produc-

tion Qi of a firm i at time t + 1 is a function of demand for its products Di and its installed

production capacity Ci: Qi(t + 1) = f(Di,Ci)(t). A trajectory of the firm’s output can be

described by logistic equation identical to (3) except that Qi replaces now Di as a depend-

ent variable. When we sum the output of all firms in a given market j, Q Qj i

i

m* �

��

1

, the

evolution of total market output is given by standard logistic equation where M denotes

the maximum market potential at a given constraint of budget, preferences, and

technology.

Q t r Q t M Q tj j j j j

* * *( ) ( ) ( )� � �1 (4)

880 Igor Matutinovic

After reaching the asymptote of the S-curve, the aggregate output of firms may fluctuate

around the maximum market potential, it may decline if Mj declines, or it can embark on

a new escalation when firms introduce innovations (figure 3).

Logistic escalation (Coren 1995) marks the phase transition in the evolution of

markets and may be triggered with a technological or marketing innovation that is usu-

ally put forth by a single firm and results in the expansion of total demand, for example,

the transition from mainframe information processing to personal computing in firms,

the introduction of GSM technology in the telephone market, and the introduction of

the “html” interface and Netscape browser in the usage of the Internet. All of these tran-

sitions opened up new market potentials and thus made possible logistic escalation of

demand.

Size Distribution of Firms

When discussing the evolution of markets we cannot skip the question of why firm

sizes are power-law distributed (highly skewed). This topic of research has a long history,

and the explanations range from stochastic processes related to growth of firms and

referring primarily to the so-called Gibrat’s law (Simon and Bonini 1958; Collins and

Preston 1961; Mansfield 1962; Quandt 1966; Dunne and Huges 1994; Hart and

Oulton 1996), entry of new firms and innovative processes (Kwasnicki 1998), and the

influence of institutional factors (Henrekson and Johansson 1998) to the effect of free

riding of employees with different preferences for leisure and income (Axtell 1999), to

name some of them. I will propose here an explanation which relates entry time,

autocatalysis, and the business strategy of the firm.

A firm’s business activity is constrained by the specific kind of knowledge embed-

ded in its employees’ skills and production technology and by the capital it is able to

raise to support its operations. Depending on these initial endowments, chosen entry

time, business strategy, and competitive environment it encounters at the time of entry,

Microeconomic Foundations of Business Cycle 881

N

t

Logistic

escalation

M

Figure 3. Logistic Dynamics of Market Behavior

Note: M = market potential; t = time.

the subsequent historical pathway of the firm unfolds in a constrained manner, in

which each successive developmental stage is dependent on the previous one and on the

entire firm’s history up to that moment. Most firms end up serving a particular set of

needs in a certain market (or a niche) and are usually not capable of arbitrarily changing

their position (the Nash equilibrium is the rigorous analogue of this situation). Suppose

that in the emergence stage of a new market a few firms start at about the same time.

Thanks to autocatalytic processes, any initial differences in entry time, knowledge, capi-

tal endowments, and chosen business strategy will soon lead to marked differences in

their growth rates. Firms with the strongest autocatalytic loop experience the fastest

growth and, consequently, considerably outgrow other competitors. Some of these

firms—usually the early entrants—that choose to position themselves in the center of the

market where they serve the widest range of needs are most likely to become the new

industry hubs. Attracted by perceived business opportunities, new firms will keep enter-

ing throughout the market evolution stages and most intensively during its growth

phase. However, once a few large companies (hubs) have occupied a considerable share

of the market, other firms have no choice but to compete for the remaining market

space by using niche specialization and innovation as their main strategies.11 One may

ask why economic state space is not occupied only by large enterprises providing all

goods and services the market requires. The answer lies in the variety of goods and ser-

vices that the modern economy needs, which require management focus, technical and

technological expertise, and specific customer knowledge. So far, no single company has

been able to serve successfully a variety of different, nonoverlapping markets. On the

contrary, since the mid 1980s there has been among large firms a trend of returning to

their core businesses, pursued in downsizing and outsourcing strategies. This process

may also be the indication that the upper size of a firm is constrained by both organiza-

tional and market factors. Large firms are usually burdened with autocratic administra-

tion and inertia (Peters 1987, 20–23) while the possibility of free riding among

employees is higher than in small firms. Therefore, as companies get larger they become

increasingly complicated and expensive to manage and this may slow down their adap-

tive capability. Because of the huge burden of fixed costs and substantial requirements

for operating capital, large companies are disproportionably more expensive to operate

than the smaller ones. In other words, “life support” sources—markets, energy, and

resources—are required in ever-larger amounts as the firm grows.

One implication of the previously mentioned constraints is that only a very small

proportion of all enterprises succeeds in reaching and maintaining its size (i.e., its

dissipative structure) at the tail of the distribution. These firms are necessarily posi-

tioned in the center of the market where demand “resource” is most abundant. Most of

the SMEs and especially owner-managed firms, however, find it more affordable to

exploit a niche and keep relatively compact, thus being more flexible to adapt and sur-

vive in a changing business environment.12 The smaller the firm, the lower its operating

costs and the initial investments required to enter a market. Therefore, whenever entry

costs are low there will be a large number of small firms (those with less than fifteen

882 Igor Matutinovic

employees13) creating and filling numerous niches in a market. Markets with low entry

costs are usually associated with the service sector of economy, which is dominating

industrialized economies in terms of GDP contribution and in the share of active

workforce. In such an economic environment, where (a) a few large firms positioned in

the center constrain the remaining market space for competitors,14 (b) low entry/operat-

ing costs associated with niche exploitation favor small firms’ size, and (c) a large service

sector provides a lot of business opportunity for small firms, it is reasonable to expect

that firm size distribution will be highly skewed at the economy level. This distribution

proves to be stable across different economies and is known to have been around for at

least fifty years in at least one economy—the USA (Simon and Bonini 1958). In terms of

self-organization, it is probably the most efficient large-scale organizational structure

that may have evolved to exploit available market opportunities in an economy.

Emergence of Productive Overcapacity

During the market growth phase, new competitors enter the market and fuel the

process of product development and niche differentiation. This process, supported by

the aggressive marketing of competing firms, creates strong stimuli for demand, which

soon grows faster then the existing productive capacities can meet. Therefore, estab-

lished competitors and potential new entrants face delicate decisions regarding invest-

ments in new productive capacities: they must estimate how much new capacity is

needed before the market is saturated and which is the proper timing for the investment

to capture more market share.15 It is a tough question given the volatility of consumer

preferences and imperfect information about the present utilization of plants available

to decision makers in competing firms (substantive and procedural uncertainty, respec-

tively). Whatever the estimated level of unsatisfied demand, the minimum investment is

constrained by chosen production technology, which simply means that production

cannot be added continuously. The investment decision is therefore constrained by pro-

duction technology at the lower end and by the business strategy of a firm at the upper

end (how much market share a particular firm aims to win from its competitors). In an

institutional and competitive environment, which requests and rewards business

growth, managers and entrepreneurs will usually take risks and actively struggle to win

market share from their competitors. This is especially true when firms experience very

intense autocatalytic growth and they do not want to lose the impetus for further

growth. The decision of a firm to expand its productive capacity is a likely outcome of

the triadic problem-solving process where the current problem of satisfying demand

growth was created by its earlier marketing actions (which at the time were just solutions

to yet another competition problem). The system-level result of many individual invest-

ment decisions (or solutions to the common competition problems) is intermittent pro-

ductive overcapacity, or overshoots, at the industry level. In other words, industry

overcapacity is an unintended consequence of competitive strategies of many individual

Microeconomic Foundations of Business Cycles 883

firms that appears at the higher hierarchical level and then feeds back on the firms’

behavior in the next period.

Capacity utilization of the firm can be expressed as an index, c C t Ci i i� ( ) / , where

Ci(t) is its capacity utilization at time t and Ci denotes maximum technical capacity of the

capital and labor of a production facility. A situation where c i= 1 denotes a threshold

where the production cost curve switches from a convex to a nonconvex regimen

because marginal product �Q can be produced only with investment in new plant capac-

ity. As long as the existing production capacity of a firm can match current demand for

its products augmented by the firm’s estimate of demand growth, �, that may arise as a

consequence of the autocatalytic process in the next relevant period t + n, Ci �Di + �(t +

1), the firm will not decide to invest in a new capacity. However, when the situation

changes to Ci �Di + �(t + 1) the firm will experience strong stimuli to expand its produc-

tive capacity. Under the constraints of nonconvex production costs and uncertainty,

such a local adaptation of individual firms to their competitive environment will inter-

mittently create gaps between total market potential and aggregate productive capacity

of a particular market, C Mij j

i

n

���

1

. Overcapacity may arise at any time during the mar-

ket growth phase as a temporary phenomenon. However, it is most likely to happen and

to persist when aggregate production approaches total market potential at the asymp-

tote of the logistic curve, marking the phase transition from a growing to the mature

market.

Emergence of an Economy Downturn

Overcapacity is created during the growth phase, but its effects on the industry are

not substantially felt before the mature phase is reached. The mature market phase is far

from any notion of equilibrium. It is characterized by a dynamic interplay of fragmenta-

tion and consolidation process where the former refers to the market splitting into finer

segments and the latter means that the introduction of a new product or service by one

firm may have such a strong market appeal that it will bring about reshuffling in the

extant distribution of market shares among competitors (Kotler 1997, 366). Demand

saturation and excess capacity intensifies competition among firms and brings about

change in the focus of competition—from market share growth to the boosting of eco-

nomic and technical efficiency.16 Firms adapt by introducing new variants of the origi-

nal product, partitioning the market into a narrower niche, and fostering technical and

technological improvements, which usually drives costs and prices down. If product

innovations fail to engender new logistic escalation of demand and profits become

eroded by invasion of each other’s segment, firms invariably adapt by reducing capac-

ity—they close down plants, lay off employees, and introduce other measures of cost cut-

ting, thus boosting their economic efficiency.17 The industry enters its downturn phase

with an intensity that is initially proportional to the gap between overcapacity and mar-

884 Igor Matutinovic

ket potential. The effect of reducing capacity and other cost-cutting measures is trans-

mitted as a ripple downstream—from final good producers to their suppliers and to

suppliers of their suppliers—or, to use Michael Porter’s terminology, to their supporting

and related industries (1990). Because of particular characteristics of a flow network

(power-law distribution of links and their flow strength, small-world effect, community

structure, and hubs), the ripple propagates throughout the economy in a nonuniform,

coordinated way. The spillover strength depends on the number and intensity of links

established among firms operating in diverse but technologically interrelated markets

and industries. Disturbances originating in different markets or at different times in the

same markets will have different effects on the economy as a whole, depending also on

the state of other variables like changes in real household income, monetary policy,

stock market variations, and world trade dynamics. Recessions will not necessarily result

every time one or more markets enter a downturn phase. However, if a hub industry

undergoes a deep and prolonged downturn the chances are that it may pull other,

related industries and eventually the whole economy into recession. Plant closure, cost

cutting, and industry-wide restructuring through mergers and acquisitions as well as

some marketing and technological innovations that take place in the downturn phase

are the expected response of firms to the overcapacity created in the boom phase. They

free the business potential for the next upswing, and the whole process is repeated again

in a vein similar to that of Schumpeter’s creative destruction (1976).

Recapitulation

The origin and propagation of upturn and downturn cycles of economic activity are

summarized in following propositions:

1. Competitive and cooperative interactions arising among firms and between

consumer good manufacturers and their respective category markets

(household demand) create constant tensions that resolve and rise again

endlessly in a triadic problem-solving process. This process, which provides

basic dynamics to economic activities, does not reach an equilibrium state as far

as there is enough low entropy in the environment to fuel the process (I refer

here to natural resources and energy).

2. In this dynamic problem-solving process, change and novelty constantly arise at

the level of individual firms. Any such change that happens at the level of a

single firm is felt as a disturbance in its environment.

3. Disturbances spread through the economy network subject to systemic flow

constraints: small-world effect and power-law distribution of the number of

trading links and their interaction strength (equations (1) and (2)).

4. Disturbances related to changes in output that arise at the level of small firms

(that comprise the majority of active establishments in every economy) tend to

die out soon and do not spread throughout the network. However,

Microeconomic Foundations of Business Cycles 885

disturbances related to innovations or new product ideas that arise at the level

of small- and medium-size firms may mark the opening of a new market and

therefore may have far-reaching implications for an industry or for the entire

economy.18 They propagate throughout the network with the growth of the

innovative firm.

5. Disturbances related to the changes in output that originate at hub firms do

spread out in a nonlinear way, reaching as far in the economy network as the

small-world effect allows it. They affect output and employment in other firms

inside as well as outside their original industry. Changes in employment usually

influence consumer confidence and therefore the spending (demand) of

households (here referred to as category markets). Similar causal reasoning

applies to large changes in government spending.

6. Since the times of the Industrial Revolution, market economies have

experienced long-term growth interrupted with intermittent recessions. The

mechanics of growth is based on autocatalytic processes of value creation

among firms in a supply chain and between firms and their respective category

markets. Growth is not constant but happens in waves of different intensity and

duration, which are often associated with technological and marketing

innovations. Beside innovations, sustained growth has been possible by

continuous depletion of natural resources (dissipation of low-entropy source).

7. Although market economies have been growing in the long run, the

medium-term potential of any individual market is always finite (i.e., a single

market evolution follows logistics trajectory). Firms are strongly motivated to

grow and gain market share in all market phases, and this often translates into

new capacity investments. Because firms are constrained in their foresight and

because production capacity cannot be added continuously (nonconvex

production costs) to meet expected demand, overcapacity gaps eventually occur

at the industry level as an unintended consequence of many individual

decisions—they are intrinsic to the competitive process itself.

8. Firms usually respond to overcapacity problems with output reduction and

cost-cutting policies—mostly by closing down plants and laying off employees.

This is an institutional arrangement which is tightly bound to another one—the

imperative of profit making in private, professionally managed firms. We may

expect cost-cutting disturbance to happen at least in one single industry at any

given time.

9. Thanks to the systemic properties of the economy network (power-law

distribution in the size of firms and in the number and intensity of their trading

links), these discrete output-reducing and cost-cutting signals are processed in a

spatially and temporally coordinated way throughout the economy, creating

intermittent downturns of all sizes and durations.

10. Therefore, at the aggregate level we expect to find a nonrandom and coherent

statistical pattern in time series of the variables usually associated with business

886 Igor Matutinovic

cycles. Because of intermittent capacity overshoots, we should find

considerable fluctuations in capacity utilization in the industry. Finally, we

should find some correspondence between peaks and troughs in capacity

utilization and dated peaks and troughs of business cycles.

Instabilities and large-scale fluctuations are thus inherent in the workings of com-

petitive markets. They do not represent a “market failure” as is conventionally assumed

by mainstream economists. Claims, occasionally expressed by mainstream economists

during a prolonged upturn, that business cycles are gone are considered to be false and,

according to this theory, represent an impossible event in the sense of Popper (1959).

However, the driving mechanism of a business cycle as presented here is expected to be

valid only under the conditions of sustained economic growth. Once this historic

growth eventually halts, due to constraints of natural resources and energy (e.g., transi-

tion to the post-oil civilization), the interactions among economic agents will probably

undergo qualitative phase transition and novel emergent phenomena will require new

explanations.

Empirical Findings

First, we will look at fluctuations in capacity utilization in the U.S. economy. Indus-

trial capacity in the U.S. economy grew 2.9 times in the observed period 1967–2003,

while capacity utilization was fluctuating within the upper and lower bounds of 89 per-

cent and 70 percent respectively (figure 4). Absolute troughs in total capacity utilization

in figure 4 indicate coordinated adaptive response (reduction of output and cost-cutting

policy) of firms in many different markets in the U.S. industry to the intermittent cre-

ation of overcapacity. Note that in their adaptive response firms have another (theoreti-

cal) choice—to clear the market at the given capacity level by reducing prices. If that

happened, capacity utilization would have fluctuated close to its theoretic maximum.

Total capacity utilization, however, never reached the 100 percent level except in war

times (Federal Reserve Bulletin 2002), and the January 1967 peak of 89 percent was not

repeated again.

A closer look at a sample of industrial sectors, which covers a wide range of techno-

logical sophistication and competitive environments, shows considerable variation in

capacity utilization at the level of a single sector and among the sectors themselves (table

1). The oil and gas industry, which is an oligopoly market and lacks most of the

autocatalytic interactions that usually lead to capacity overshoot as described in the pre-

ceding section, shows the smallest variability in capacity utilization. Its statistics on

capacity utilization are therefore taken as a benchmark for assessment of the degree of

capacity utilization variability in other sectors. Row 6 shows standard deviation (�) of

each sector expressed as a percentage of the � in oil and gas industry. We see, for exam-

Microeconomic Foundations of Business Cycles 887

ple, that fluctuations in capacity utilization in the transport equipment sector are three

times and in the iron and steel sector six times greater than in the oil and gas sector. We

also note that most of the sectors in the sample have troughs below the industry average

of 70,90 percent. The Kolgomorov-Smirnov test in column 7 shows that fluctuations in

the total industry and in most of the sectors are not Gaussian, which is consistent with

the claim that they arise as a consequence of coordinated interactions among agents and

are not just a “noise” in the market dynamics. Observed fluctuations in capacity utiliza-

tion do not invalidate the prediction arising from overcapacity theory (see item 10 in the

previous section).

Coordinated interaction of agents may be responsible for another important phe-

nomenon of business cycles—that of industry comovement (Moore 1993; Hornstein

2000). Andreas Hornstein showed that, over the business cycle, activity in almost all

industries of the economy simultaneously increases and decreases. This comovement

can be observed for a wide variety of activity measures, such as gross output, value

added, employment, the use of capital services, or intermediate inputs. After listing vari-

ous possible causes, Hornstein concluded that the question of industry comovement

has not been addressed successfully (2000). Overcapacity theory offers a simple and

straightforward explanation for this phenomenon: it arises from interactions of func-

tionally related firms, which are interconnected on the network with a small-world

effect, so that changes in output in one industry occasionally span all industries (see the

section on the emergence of industry downturn).

888 Igor Matutinovic

70

72

74

76

78

80

82

84

86

88

90

1967

1968

1970

1971

1973

1974

1976

1977

1979

1980

1982

1983

1985

1986

1988

1989

1991

1992

1994

1995

1997

1998

2000

2001

2003

Ca

pacity u

tiliz

ation

in

%

Figure 4. Capacity Utilization in U.S. Industry 1967–2003

Notes: The picture shows monthly capacity utilization in percentages. The rate of capacity utilization equals the seasonally ad-

justed output index expressed as a percentage of the related capacity index. Data on capacity and utilization: Economic Time

Series page, http://www.economagic.com. Detailed explanations of terms are at Federal Reserve site, http://

www.federalreserve.gov/releases/G17.

Although peaks and troughs of capacity utilization follow their own aperiodic

dynamics, they bear a general resemblance to peaks and troughs of business cycles as

dated by Christina Romer (1999) and by NBER (2003). The analysis of monthly data on

capacity utilization shows that it reaches its maximum from several months to several

years before the upturn peak as dated by NBER (2003). However, half of capacity utiliza-

tion periodic minimums coincide with NBER’s dated troughs (figure 5), and capacity

utilization at dated peaks is still very high (above 80 percent except for March 2001).

This conforms to the prediction about the existence of certain correspondence between

peaks and troughs in capacity utilization and dated peaks and troughs of business

cycles.19

In an economic environment dominated by international trade, the extent of over-

capacity that arises in any industry is best visible at the global level as manufacturers usu-

ally base their investment in capacity on both domestic and international demand. The

United Nations 2003 annual economic survey emphasized the negative impacts of over-

capacity created by excessive investments in the 1990s, especially in the IT sector. Exam-

ples taken from the business literature in the period 2001–2003, which comprises the

last recession in the USA and a prolonged recession in various industries worldwide,

provides an insight into the extent of overcapacity. Reliable sources of business informa-

Microeconomic Foundations of Business Cycles 889

Table 1. Fluctuations in Capacity Utilization in U.S. Industry

1 2 3 4 5 6 7

Industry N Range Minimum Mean � �(oil) = 100 K-S test

Oil and gas 386 10,54 85,65 91,98 1,99 1,00 0,01

Food, tobacco, and

beverages

446 10,24 76,34 82,41 2,04 1,02 0,04

Apparel and leather 386 27,92 61,26 79,49 5,59 2,80 0,00

Chemical 674 22,55 68,05 79,64 4,34 2,18 0,16

Textiles 386 30,42 62,70 83,24 5,59 2,80 0,02

Transport 446 35,07 55,25 76,31 6,37 3,19 0,07

Wood 386 35,67 59,58 80,22 6,47 3,25 0,00

Plastics 674 37,25 61,59 85,02 6,60 3,31 0,00

Electric equip. 386 34,68 64,48 82,93 6,67 3,35 0,21

Computers and

peripherals

386 32,90 60,88 78,13 6,83 3,43 0,16

Furniture 446 34,33 64,20 80,39 7,20 3,61 0,01

Machinery 446 36,61 57,69 79,92 7,91 3,97 0,01

Automobile 446 36,61 57,69 79,92 7,91 3,97 0,01

Steel 386 67,29 37,76 81,13 12,48 6,26 0,00

Total industry 446 18,49 70,90 81,55 3,80 1,91 0,05

Source: Economagic.com.

Note: Data cover three initial periods: N(674) = 1948; N(446) = 1967; N(386) = 1972, and all end on February 2004.

tion like Financial Times and business writers (Greider 1997) report evidence of global

overcapacity in a number of important industries like semiconductors, computers, con-

sumer electronics, commercial aircraft, textiles, steel, automotive, and pharmaceutical.

Note that subsidies, which governments extend to their exporters, contribute to cre-

ation of excessive capacity at the world level (take, for example, the steel industry and the

recent trade dispute at WTO between the USA and major steel-exporting countries).

Overcapacity may coincide with a substantial fall in demand when a market

becomes saturated in its maturity stage. A recent case of such a coordinated state of over-

capacity and overall demand drop in interrelated markets—personal computing, con-

sumer electronics, communication equipment, and semiconductors—happened during

the 2001 recession.20 Demand momentum, which was primarily created by aggressive

marketing, transferred eventually into new capacity investments (temporary increase in

the autocatalytic level). Once the market became saturated the ensuing drop in demand

necessarily coincided with overcapacity and unsold inventories. During the recession

period the whole autocatalytic activity in the industry settled at a new, lower level thanks

to the adaptive process of plant closure and massive lay-offs of the workforce.

Even when an industry faces overcapacity for several years, like the automotive one,

individual manufacturers may decide to increase their capacity because they have for-

ward product plans that assume that they will gain market share.21 This is consistent

890 Igor Matutinovic

85,33

77,57

88,49

74,49

84,42

77,73

80,96

71,48

82,82

78,62 79,05

75,09

70

72

74

76

78

80

82

84

86

88

90

1969

Dec

1970

Nov

1973

Nov

1975

Mar

1980

Jan

1980

July

1981

July

1982

Nov

1990

July

1991

Mar

2001

Mar

2001

Nov

Figure 5. Capacity Utilization in U.S. Industry 1967–2001 on Dated Peaks and

Troughs

Sources: Dates of peaks and troughs are from NBER; capacity utilization is from Economagic.com.

Notes: Capacity utilization recorded at the NBER dated troughs coincide with its absolute minimums (marked with ovals) in

that period. Peaks of capacity utilization always precede the NBER dated boom peaks except for the peak July 1981, where

they coincide.

with the earlier claim that managers and entrepreneurs willingly take risks and actively

struggle to win market share from their competitors. All these examples show that the

mechanism of overshoot and subsequent contraction in business activity is intrinsic to

competitive interactions among firms and to marketing interactions that producers of

final goods engender with households. In a forthcoming work (Matutinovic 2006) I

present empirical evidence of power-law distributions in several variables that describe

business cycles in the U.S. economy (duration of recessions and recoveries, severity of

recessions, changes in industrial output, inventories, and employment). This evidence

is congruent with the prediction expressed in item 10, namely, that statistical descrip-

tion of variables describing business cycles should be coherent and non-Gaussian. Simi-

lar results are also reported in Ormerod 2002.

Discussion and Conclusions

The present description of business cycle mechanisms departs radically from main-

stream theorizing in that field—that of real business cycles (RBC)—in at least two impor-

tant aspects: the first and the self-evident one is that an economy does not need any

external shock to drive its large-scale fluctuations. The dynamics is self-generating and

ingrained in agents’ behavior and their institutional environment. Second, the empha-

sis is shifted from random processes governing economic interactions to functional

coordination and path dependency inherent in the very structure of economic network.

The role of random shocks in economic dynamics deserves a few lines more. As early as

fifty years ago John Hicks (1950) observed that to explain fluctuations by means of ran-

dom shocks is tantamount to confessions of ignorance (quoted in Medio 1992, 19). In

the literature these shocks are usually attributed to demand and technology. Demand

shocks can come from government policy and international trade (Dow 1998) or from

the consumer sector (Bak et al. 1991; Schneikman and Woodford 1994). However, as I

have tried to demonstrate here, supply and demand are so intrinsically intertwined that

referring to “demand shocks” as exogenous to competitive interactions among firms

falls short of providing plausible explanation. Victor Zarnowitz (1999) made a similar

observation in his critique of monocausal origins of business cycles and argued that

cyclical boom-and-bust imbalances always refer to the interplay of two market sides.

The overcapacity mechanism presented here is valid for generating recessions of all

sizes. This can be contrasted with the work of Christopher Dow (1998), who claimed

that large recessions require different explanations from the small ones and focused his

theory on the interplay between exogenous shocks and changes in business and con-

sumer confidence. There is evidence (Ormerod 2002; Matutinovic 2006) that reces-

sions do come in a scale-free or power-law pattern, at least in the U.S. economy, and this

calls for a theory that can provide a comprehensive explanation.

Although government appears as a hub in an economy network and therefore its

spending may have far-reaching implications, this work does not support the idea that

Microeconomic Foundations of Business Cycles 891

government policies may be the primary cause of depressions in the twentieth century

(Kehoe and Prescott 2002; Bergoeing et al. 2002). The fluctuations generated by com-

petitive dynamics of firms suffice to induce both booms and boosts in the economy.

However, government policies may influence the intensity of both expansions and

recessions, and in that sense I believe that monetary and fiscal policies contributed to

moderating the size of postwar recessions in the industrialized world. Given the unpre-

dictability of economic dynamics, which is characteristic of complex adaptive systems in

general, the present toolbox of countercyclical policies, which developed from the Great

Depression onward, probably represents the limit of rational intervention in a market

economy.

The main difference between the present approach and older endogenous theoriz-

ing, like trade cycle theories, is in the hierarchical level at which the events are

explained. Conventionally, the origin of imbalance between productive capacity and

consumer demand has been explained at the aggregate level, usually contraposing sav-

ings and investment or wages and profits. Some of the aggregate models, like the ones

proposed by Zarnowitz (1999) and John Harvey (2002), provide a valuable insight into

the relationship between macro variables and the contingency in their fluctuations.

Harvey’s (2002) model of a Keynes-style trade cycle shows how interplay between mar-

ginal efficiency of capital (which is to an extent a function of capacity utilization) and

expectations on future profits contribute to the reversal of upswing and downswing

phases. The crisis of expectations, and the subsequent recession, occurs whenever there

is a persisting positive gap between expected (speculative) and real (objective) yields on

capital (Harvey 2002). In other words, when profits in an industry continue to fall for

several consecutive periods, then expectations about its improvement vanish and man-

agement usually decides on cost-cutting measures to restore them. The bottom-up the-

ory of market dynamics and overcapacity creation presented here abridges the gap

between macroeconomic flows described in Harvey’s (2002) and similar models and the

microeconomic behavior that underlies it.

Autocatalytic loops that form among firms and their markets at the microeconomic

level create corresponding dynamics at the aggregate level. This dynamics is generally

well captured by the accelerator-multiplier model, conceived independently by Keynes

and Michal Kalecki (Sherman 1991; Benic 2002). Firms actively stimulate demand in

the household sector, and this feeds back via accelerator into new investments and

employment.22 The income thus earned transfers partially into consumer demand via

the multiplier, and therefore the circle of cumulative, self-reinforcing causation of

expansion is closed. In the multiplier-accelerator model, the onset of a downturn phase

is identified in the decreasing share of labor in national income during the expansion

phase, which results in falling of consumer demand and, therefore, in the reversal of the

cumulative causation loop (Sherman 1991, 151, 200–207). Under the present theory,

the main cause of a downturn resides in the combined effects of production overcapac-

ity created by occasional investment overshoots and subsequent cost-cutting policies as a

response to falling corporate profits. These causes may combine with a decreasing share

892 Igor Matutinovic

of labor in national income and thus jointly contribute to the reversal of the expansion

phase.

The importance of self–reinforcing loops of business activity in business cycles gen-

eration has already been emphasized in Mitchell’s works. Autocatalytic loops that form

among firms and their markets, under the constraints of institutions, uncertainty, and

production indivisibilities, necessarily produce coordinated fluctuations in output,

capacity utilization, employment, and other variables that mark business cycles. To the

extent that this mechanism is true, downturns are inscribed in the very functioning of

market economies. This is the key conclusion of the paper. It contrasts, however, with

some endogenous theories, like, for example, that of Zarnowitz, who in his seminal

book (1992, xvi) stated that his theorizing on business cycles does not imply that con-

tractions in general are inevitable or must recur with any frequency.

The mechanism generating autocatalytic growth and, eventually, overcapacity and

occasional recessions is not the sole cause of business cycle dynamics, but it represents

its essential and sufficient microeconomic driver. It lies at the very core of the competi-

tive game played in capitalist economies. Other factors may interfere with this basic

mechanism, for example, changes in energy prices, stock market speculations, govern-

ment fiscal and monetary policy, or international political (in)stability, but they do not

generate the overall process of large-scale fluctuations—they mostly reinforce or damp

the underlying microdynamics. The theory presented here is “historically specific” in

terms of Sherman (2001) as it shows how one specific set of institutions, characteristic

of a capitalist economy—profit-making imperative, pursuit of growth by techno-

structure, and widely shared cost-cutting policy in times of falling profits—gives rise to

the phenomenon of business cycles. In a socialist economy, which lacks these institu-

tions, the unfolding of large-scale fluctuations should be significantly different.23 Simi-

lar to Sherman (1991, 2003), but from a different and complementary perspective, this

work supports the view that recessions on national and global levels cannot be

avoided—they are here to stay with us as an irremovable systemic feature of the capitalist

economy. There can be no “new economy” (De Long 1999) or “Golden Age” (Zarnowitz

1999) that could free us from downturns in the future, large depressions included.

Notes

1. This initially very influential theoretical branch was pushed aside in the postwar period after

the influential works of R. Frisch and E. Slutsky, who in the 1930s argued that random shocks

and propagation mechanisms acting on a fundamentally stable economy provided a better

account of business cycles (Dow 1998; Blanchard 2000). This approach culminated in real

business cycle models, which still constitute the majority of effort in the field.

2. This view was especially outlined in the theories of Atkinskon Hobson and Mikail Ivanovic

Tugan-Baranovskij (Screpanti and Zamagni 1993, 219–220). Karl Marx, on the other hand,

argued that a discrepancy between the growth of investments and the growth of wages that

unfolds in the upturn phase fails to create adequate demand for the goods produced, and

Microeconomic Foundations of Business Cycles 893

investments eventually create overcapacity and stocks of unsold goods that must be cleared in

the downturn phase of the cycle (Fusfeld 1972, 69–70).

3. The concept of small world, also known as “six degrees of separation,” dates back to sociologist

Stanley Milgram, who in 1967 showed experimentally that any two persons in the USA are

separated by only 5.5 intermediate persons.

4. In random networks, characteristic path length increases logarithmically with the size of net-

work (with the addition of new nodes).

5. More precisely, I expect that empirical observations will find broad-scale or truncated scale-free

networks (i.e., there is a finite size effect in the connectivity pattern), P(ki) ~ ki–� �(k/�), where

�(k/�) introduces a cutoff at some characteristic scale � (Solé et al. 2003). As this detail does not

have any impact on the system behavior I will disregard it.

6. J. J. Ramsden and Gy Kiss-Haypál (2000) analyzed twenty countries in America, Asia, and

Europe. Their analysis, however, includes only several hundred to several thousand largest

companies measured by annual revenue and leaves out the majority of the firm’s population. I

calculated power-law distribution of firm sizes in Sweden (directly from data) and indirectly

for twelve other West European countries (from the relative proportions of small, medium,

and large firms; data are reported in Henrekson and Johansson 1998), for Hong Kong directly

from published data (Hong Kong MDS 1996), and for Croatia from the Croatian Chamber of

Commerce 2000 database. In all cases, the employees number criterion for size was used. The

fact that we encounter the same pattern in different economies across the world provides a

reasonable clue that this pattern might be universal indeed.

7. An exception is the banking sector, which shows considerable clustering because of interbank

lending practice (see Boss et al. 2003).

8. What is “exogenous” in open systems depends exclusively on the selection of the domain of

system description or the arbitrary decision of where to place system boundaries.

9. It can be easily verified by setting C = 1, Q(t = 1) � 0,001 and then varying parameters r, a, and

b. Logistic dynamics is best observed for values of r � 1.20.

10. It is a well-established empirical fact that 60 percent to 80 percent of newly established firms

go bankrupt within the first few years of activity (Kennedy 1985).

11. Specialization in the economy may play the same functional role as niche separation and com-

petitive exclusion in biology.

12. For example, field studies conducted in Sweden among small business owners and managers

point to their reluctance to foster growth, for reasons related to costs and risks involved with

business expansion (Henrekson and Johansson 1998). Their reluctance, was, however,

related to an institutional environment which they perceived as unfavorable for pursuing

growth.

13. In the U.S. economy, for example, 82 percent of all firms are found in the size classes between

one and thirteen employees (data source in Axtell 2001).

14. Generally, a rule of thumb of 20:80 also applies for market shares in any given industry, where

approximately 20 percent of large firms hold 80 percent of market.

15. Firms know that failing to grow with the market equals losing market share, an inconvenience

that usually results in lower profits and exposure to takeovers.

16. I am thankful to Robert Ulanowicz for pointing out to me the importance of the strategic shift

from growth to efficiency in the different development phases of ecosystems, which has its

analogy in the evolution of markets.

17. The extent of cost-cutting measures across industries can be easily grasped with occasional

inspection of business newspapers and literature.

18. Among numerous examples of successful product and marketing innovations that arose in

small firms and sent ripples of change throughout their industry, there are Microsoft with its

DOS operating system that revolutionized the usage of computers in business and house-

holds; Kodak in the nonprofessional camera market, and Kentucky Fried Chicken in the

fast-food market. Consider also the recent case of Google, which was started by two students

894 Igor Matutinovic

and in only six years reached $962 million in revenues, challenging more established competi-

tors like Microsoft and Yahoo!

19. This correspondence cannot be fully established anyway because the NBER Business Cycle

Dating Committee does not include capacity utilization among the relevant indicators and

because major indicators usually peak or trough in different months; therefore, judgment is

decisive in the dating procedure (for details see www.nber.org).

20. Financial Times, “The Chips Are Down, But Hope Still Flickers,” April 17, 2002.

21. Research World, “Keeping a Lid on Overcapacity,” ESOMAR, January 2004.

22. Because of production indivisibilities, during the expansion phase the accelerator plays a non-

linear role in transferring demand for consumer goods into new investments. In that sense,

standard equations of the accelerator-multiplier models (see Sherman 1991, 153–154) are

not adequate. Similarly, the disinvestment of capital in the downturn phase does not stop at

the level of depreciation, as it is generally assumed, but goes beyond it as a consequence of

cost-cutting policies and shutting down of redundant production sites.

23. It is certain that institutions of profit and cost cutting are absent in a communist economy

(e.g., the former SSSR) and relatively weak in a socialist one (e.g., the former Yugoslavia). The

pursuit of growth by technostructure, however, was not irrelevant in a socialist system but

without the support of the former two institutions could not have been operative in genera-

tion of business cycles.

References

Amaral, Luis A. N., A. Scala, M. Barthélémy, and H. E. Stanley. “Classes of Small-World Networks.” Proceedings

of the National Academy of Sciences 97, no. 21 (2000): 11149–11152.

Arthur, W. Brian. “Positive Feedbacks in the Economy.” Scientific American, February (1990): 92-99.

Auyang, Y. Sunny. Foundations of Complex System Theories in Economics, Evolutionary Biology, and Statistical Physics.

Cambridge: Cambridge University Press, 1998.

Axtell, Robert. “The Emergence of Firms in a Population of Agents.” SFI working paper 99-03-019, Santa Fe In-

stitute, 1999.

———. “Zipf Distribution of U.S. Firm Sizes.” Science 293 (2001): 1818–1820.

Bak, Per, and Chen, Khan. “Self-Organized Criticality.” Scientific American 264, no. 1 (1991): 26–33.

Barabási, Albert. Linked: The New Science of Networks. Cambridge, Mass.: Perseus Publishing, 2002.

Barabási, Albert L., and Réka Albert. “Emergence of Scaling in Random Networks.” Science 286 (1999):

509–512.

Barabási, Albert L., and Eric Bonabeu. “Scale-Free Networks.” Scientific American (May 2003): 50–59.

Benic, Ðuro. “Business Cycles.” Ekonomska misao i praksa 11, no. 1 (2002): 11–67. (In Croatian).

Bergoeing, Raphael, Patrick J. Kehoe, Timothy J. Kehoe, and Raimundo Soto. “Policy-Driven Productivity in

Chile and Mexico in the 1980’s and 1990’s.” American Economic Review 92, no. 2 (2002): 16–21.

Blanchard, Olivier. “What Do We Know about Macroeconomics that Fisher and Wicksell Did Not? The Quar-

terly Journal of Economics 115 (2000): 1375–1409.

Boss, Michael, Helmut Elsinger, Martin Summer, and Stefan Thurner. “The Network Topology of the Inter-

bank Market.” SFI working paper 03-10-054. Santa Fe Institute, 2003.

Casti, L. John. Complexification: Explaining a Paradoxical World through the Science of Surprise. New York: Harper

Collins, 1994, 37–40.

Cocks, Dough. “On Deconstructing Complex Adaptive Systems.” Talk to initial meeting of Complex Adaptive

Systems Discussion Group in CSIRO Division of Wildlife and Ecology, Canberra, March 17, 1999.

http:// www.labshop.com.au/dougcocks.

Collins, Norman J., and Lee E. Preston. “The Structure of the Largest Industrial Firms: 1909–1958.” American

Economic Review (1961): 986–1003.

Coren, L. Richard. “Logistic Escalation as the Mechanism of Emergence.” Cybernetica 38, no. 3 (1995):

201–214.

Microeconomic Foundations of Business Cycles 895

Dallago, Bruno. “The Organizational Effect of the Economic System.” Journal of Economic Issues 36, no. 4

(2002): 953–979.

De Long, Bradford. “Introduction to the Symposium on Business Cycles.” Journal of Economic Perspectives 13,

no. 2 (1999): 19–22.

Dequech, David. “Bounded Rationality, Institutions, and Uncertainty.” Journal of Economic Issues 35, no. 4

(2001): 911–929.

Dorogovtsev, Serguei N., and José F. F. Mendes. “Evolution of Networks.” 2001. http://arxiv.org/abs/

cond-mat/0106144v2.

Dosi, G., and M. Egidi. “Substantive and Procedural Uncertainty.” Journal of Evolutionary Economics 1 (1991):

145–68.

Dow, Christopher. Major Recessions: Britain and the World, 1920–1995. Oxford: Oxford University Press, 1998.

Dunne, Paul, and A. Hughes. “Age, Size, Growth, and Survival: UK Companies in the 1980s.” Journal of Indus-

trial Economics 42, no. 2 (1994): 115–40.

Federal Reserve. “Industrial Production and Capacity Utilization.” Bulletin G 17. 2002. http://

www.federalreserve.gov.

Fuhrer, Jeffrey C., and Scott Schuh. “Beyond Shocks: What Causes Business Cycles? An Overview.” New Eng-

land Economic Review (November/December 1998): 3–24.

Fusfeld, R. Daniel. The Age of Economists. Glenview: Scott, Freeman and Company, 1972. Translation: Storia

del pensiero economice moderno. Torino: Arnolodo Mondadori Editore, 1976.

Fontana, Giuseppe, and Bill Gerrard. “A Post Keynesian Theory of Decision Making under Uncertainty.” Jour-

nal of Economic Psychology 25 (2004): 619–637.

Galbraith, John K. Economics and the Public Purpose. New York: Mentor, 1973.

Girvan, Michelle, and M. E. J. Newman. “Community Structure in Social and Biological Networks.” SFI work-

ing paper 01-12-077, Santa Fe Institute, 2001.

Greider, William. One World, Ready or Not: The Manic Logic of Global Capitalism. New York: Touchstone, 1997,

111–119.

Grübler, Arnulf. “Time for a Change: On the Patterns of Diffusion of Innovation.” Daedalus 125, no. 3 (1996):

19–42.

Hart, Peter E., and Nicholas Oulton. “Growth and Size of Firms.” Economic Journal 106 (September 1996):

1242–1252.

Harvey, John T. “Keynes’ Chapter Twenty-Two: A System Dynamics Model.” Journal of Economic Issues 36, no. 2

(2002): 373–381.

Henrekson, Magnus, and Dan Johansson. “Institutional Effects on the Evolution of the Size Distribution of

Firms.” Small Business Economics 12, no. 1 (1998): 11–23.

Hong Kong Monthly Digest of Statistics. Hong Kong: Census and Statistics Department, 1996.

Hornstein, Andreas. “The Business Cycle and Industry Comovement.” Economic Quarterly 86, no. 1 (2000):

27–48.

Kehoe, Timothy J., and Edward C. Prescott. “Introduction—Great Depressions of the 20th Century.” Review of

Economic Dynamics 5, no. 1 (2002): 1–18.

Kennedy, R. Carson. “Thinking of Opening Your Own Business? Be Prepared!” Business Horizons, Septem-

ber–October, 1985, 38–42.

Keynes, John Maynard. “The General Theory of Employment, Interest, and Money.” 1937. Reprinted in The

Collected Writings of John Maynard Keynes: Vol. VII. London: Macmillan for the Royal Economic Society,

1973a.

———. “The General Theory of Employment.” 1937. Reprinted in The Collected Writings of John Maynard Keynes:

Vol. XIV: The General Theory of Employment, Interest, and Money. London: Macmillan for the Royal Eco-

nomic Society, 1973b.

Kotler, Phillip. Marketing Management. Upper Sadde River, N.J.: Prentice Hall, 1997.

Kwasnicki, Witold. “Skew Distributions of Firms’ Sizes—An Evolutionary Perspective.” Structural Change and

Economic Dynamics (1998): 135–158.

Mansfield, Edwin. “Entry, Gibrait’s Law, and the Growth of Firms.” American Economic Review (1962):

1023–1051.

896 Igor Matutinovic

Marchetti, Cesare. “On a Fifty Years Pulsation in Human Affairs: Analysis of Some Physical Indicators.” IIASA

professional papers PP-83-5, 1983.

Matutinovic, Igor. “Self-Organization and Design in Capitalist Economies.” Journal of Economic Issues (forth-

coming 2006).

Medio, Alfredo. Chaotic Dynamics: Theory and Application to Economics. Cambridge: Cambridge University Press,

1992.

Milgram, Stanley. “The Small World Problem.” Psychology Today (May 1967): 60–67.

Mitchell, Wesley C. Business Cycle: The Problem and Its Setting. 1927. Reprint, New York: National Bureau of

Economic Research, 1954.

———. Business Cycles and Their Causes. Berkeley: University of California Press, 1941.

Monod, Jacques. Le hasard et la necéssité. 1970. Translation, Il caso e la necessitá. Torino: Arnoldo Mondadori

Editore, 1974.

Montoya, José M., and Richard V. Solé. “Small World Patterns in Food Web.” SFI working paper 00-10-059,

Santa Fe Institute, 2000.

Moore, Geoffrey H. “Recession.” In The Fortune Encyclopedia of Economics, edited by David R. Henderson. New

York: Warner Books, 1993.

National Bureau of Economic Research (NBER). “US Business Cycle Expansions and Contractions.” 2003.

http://release.nber.org/cycles.

Ormerod, Paul. Butterfly Economics: A New General Theory of Social and Economic Behavior. New York: Pantheon

Books, 1998.

———. “The US Business Cycle: Power Law Scaling for Interacting Units with Complex Internal Structure.”

Physica A 314 (2002): 774–785.

Padgett, John F., Lee Dowaan, and Nick Collier. “Economic Production as Chemistry.” SFI working paper

03-02-010, Santa Fe Institute, 2003.

Peters, Tom. Thriving on Chaos. New York: Harper & Row, 1987.

Popper, Karl R. The Logic of Scientific Discovery. New York and London: Routledge, 1959.

———. Knowledge and the Body-Mind Problem: In Defense of Interaction, edited by M. A. Notturno. New York and

London: Routledge, 1994.

Porter, Michael E. The Competitive Advantages of Nations. London: Macmillan Press, 1990.

Quandt, Richard E. “On the Size Distribution of Firms. American Economic Review 56 (1966): 416–432.

Ramsden, J. J., and Gy Kiss-Haypál. “Company Size Distribution in Different Countries.” Physica A 277 (2000):

220–227.

Romer, Christina R. “Changes in Business Cycles: Evidence and Explanations.” Journal of Economic Perspectives

13, no. 2 (1999): 23–44.

Rutherford, Malcolm. Institutions and Economics: The Old and New Institutionalism. Cambridge: Cambridge Uni-

versity Press, 1996.

Schneikman, José, and Michael Woodford. “Self-Organized Criticality and Economic Fluctuations.” American

Economic Review 84, no. 2 (1994): 417–421.

Schumpeter, Joseph A. Capitalism, Socialism, and Democracy. London: George Allen & Unwin, 1976. Transla-

tion in Croatian: Kapitalizam, socijalizam i demokracija. Beograd: Plato, 1998.

Screpanti, Ernesto, and Stefano Zamagni. An Outline of the History of Economic Thought. Oxford: Claredon Press,

1993.

Sherman, Howard J. The Business Cycle: Growth and Crisis under Capitalism. Princeton: Princeton University

Press, 1991.

———. “The Business Cycle Theory of Wesley Mitchell.” Journal of Economic Issues 35, no. 1 (2001): 85–97.

———. “Institutions and the Business Cycle.” Journal of Economic Issues 37, no. 3 (2003): 1–22.

Simon, Herbert A., and C. P. Bonini. “The Size Distribution of Business Firms.” American Economic Review 48

(1958): 607–617.

Solé, Richard V., Ramon Ferrer-Cancho, José M. Montoya, and Sergi Valverde. “Selection, Tinkering, and

Emergence in Complex Networks.” Complexity 8, no. 1 (2003): 20–33.

Souma, Fujiwara W. Y., and H. Aoyma. “Complex Networks and Economics.” Physica A 324 (2003): 396–401.

Microeconomic Foundations of Business Cycles 897

Stanley, Michael H. R, Luís A. N. Amaral, Sergey V. Buldyerv, Shlomo Havlin, Heiko Leschhorn, Philipp

Maass, Michael A. Salinger, and Eugene H. Stanley. “Scaling Behavior in the Growth of Companies.” Na-

ture 379 (1996): 804– 806.

Ulanowicz, Robert E. Ecology, the Ascendant Perspective. New York: Columbia University Press, 1997.

———. “Life after Newton: An Ecological Metaphysic.” BioSystems 50 (1999): 127–142.

United Nations. The World Economic and Social Survey. The United Nations, 2003.

Watts, Duncan J., and Steven H. Strogatz. “Collective Dynamics of Small-World Networks.” Nature 393 (June

4, 1998): 202–204.

Zarnowitz, Victor. Business Cycles: Theory, Indicators, and Forecasting. Chicago: The University of Chicago Press,

1992.

———. “Theory and History behind Business Cycles: Are the 1990s the Onset of a Golden Age?” Journal of Eco-

nomic Perspectives 13, no. 2 (1999): 69–90.

898 Igor Matutinovic