Stavros Ioannides

PanteionUniversity

Department of Political Science and History

136 Syngrou Avenue

176 71 Athens

Greece

Email:

And

Ioanna Pepelasis Minoglou

AthensUniversity of Economics and Business

Department of Economics

76 Patission Avenue

104 34 Athens

Greece

Email:

Explaining the longevity of market-embedded clans:

the case of Greek shipping[1]

Preliminary draft

May 2006

Abstract

Most attempts to explain the superior performance of ethnic networks that operate in international markets describe these networks as static institutions. Thus, they cannot account for instances in which an ethnic institution may itself evolve and display a capacity to adapt to changing circumstances. This paper builds a theoretical framework that can address such instances, taking the example of Greek shipping as a case in point. The fact that we are seeking to describe a dynamic institution prompts us to turn to evolutionary economics. Thus, the institution we will be describing must have the capacity to change over time, while it must have the unintended consequence of sustaining the group’s competitive advantage. On these grounds, we construct the theoretical concept of the market-embedded clan. We argue that market-embedded clans have important implications for the business forms that members tend to set up. The key conclusion of our analysis is that the efficiency of market-embedded clans versus other forms of business networks stems not only from their capacity to spread risk, but most importantly, from the fact that they ensure members superior access to entrepreneurial opportunities than non-members.

JEL Classification: L22;N83;N84;Z13.

Keywords: Evolutionary Economics; Business Organizations; Ethnic Networks; Clans; Shipping.

1Introduction

There are instances where a network of agents supplying a market, outperform other agents, who serve the same market but are not members of the specific group, despite the fact that the characteristics of the membership do not warrant such apparent differences in performance. Such instances have been identified and discussed extensively in the literature on ethnic networks. Examples include the Jewish domination of the diamond trade in Antwerp and New York, Chinese merchants in Southeast Asia, Korean grocers in some major American cities, and many others. In all these instances, while the strong correlation between superior performance and ethnic origin is beyond doubt, the causal link between the two is far from obvious.

Most attempts to explain this puzzle describe ethnic networks as static institutions, which are perceived either as having emerged and being preserved through non-economic means, or as deliberately created constructs that economize on transaction and information costs. However, important as these insights may be, they cannot account for instances in which the superior performance of an ethnic network is sustained for long periods of time. In such cases, the institution that ensures the group’s superior performance may itself evolve and display a capacity to adapt to changing circumstances.

A good example of the longevity of an ethnic institution is Greek-owned shipping. For almost two centuries Greek-owned shipping has commanded market shares that are entirely inexplicable by reference to Greek entrepreneurial success in any other branch of business. How can this phenomenon be explained? This paper attempts to build a theoretical framework that can satisfactorily address such instances, taking the example of Greek shipping as a case in point.

The fact that we are seeking to describe a dynamic institution, i.e. an institution that evolves through time, prompts us to turn to evolutionary economics for ideas. We take heed from Loasby’s (1999, p. 105) contention that ‘institutional economics must be evolutionary economics…. and evolutionary economics must be institutional economics’. The essence of the evolutionary conception of the economy is that economic phenomena must be conceived as processes. On these grounds Loasby (ibid) maintains that ‘institutions are a response to incomplete knowledge… they may have unexpected consequences… and are likely to change over time’. Thus, the institution we will be describing must have the capacity to change over time, while it must have the unintended consequence of sustaining the group’s competitive advantage through time. On these grounds, this paper attempts to construct the theoretical concept of the market-embedded clan, which is a further development of the clan concept originally formulated by William Ouchi in 1980.

Ouchi put forth the concept of the clan as a potential alternative to markets and hierarchies, the polar organizational forms for Transaction Costs Economics. A clan is a group of individuals characterized by high degrees of ‘equity’, ‘goal congruence’, and shared understanding. These attributes allow members to act in a coordinated fashion with minimal levels of bureaucratic control. Therefore, Ouchi’s original concept refers to a group that belongs to an organization, or is the organization. Thus the clan is thought of as a substitute for hierarchy.

However, the ethnic networks we want to deal with in this paper consist of independent traders, who do not act for a common goal but pursue their own selfish interests. Therefore, we attempt here to extend the theory of the clan, by analyzing its relevance in the context of arm’s-length market relations. Thus, we have constructed in Pepelasis Minoglou and Ioannides (2004) the concept of the market-embedded clan, which we contrast to the ‘organization-embedded clan’, as we describe Ouchi’s original concept. In that paper we argued that market-embedded clans have important implications for the business forms that members tend to set up. The key conclusion of our analysis is that the efficiency of market-embedded clans versus other forms of business networks stems not only from their capacity to spread risk, but most importantly, from the fact that they ensure members superior access to entrepreneurial opportunities than non-members. In this study we develop the evolutionary implications of the concept, as we seek to address the longevity of an institution: Greek shipping over two centuries.

The paper is organized as follows. Section 2 discusses the problems related with institutional longevity that two widely cited accounts of ethnic trading networks, which are based on the static ideas of mainstream economics, have faced. Section 3 presents a simple model that aims at the formal description of these problems. The evolutionary ideas on which an alternative view of ethnic trading networks can be based are introduced in Section 4. Section 5 discusses William Ouchi’s original formulation of the ‘clan’ concept, while in Section 6 we attempt to extend the original concept to our concept of ‘market-embedded’ clans. In Section 7, we argue that the attributes of market-embedded clans may indeed account for their capacity to survive and evolve through significant periods of time. We demonstrate the explanatory relevance of our analysis with data drawn from the historical evolution of the Greek shipping community, arguing that its longevity can be attributed to the fact that it has always displayed a set of market-embedded-clan-like characteristics. Finally, we sum up our conclusions in Section 8.

2Ethnic networks in non-evolutionary perspectives

All accounts of the superior performance of ethnic networks –including the one we put forth in this paper place emphasis on the high intensity of trust relations among their members. Trust is the reason that members prefer to cooperate with members rather than with non-members. The accounts that are based on mainstream economics view the institutions that sustain trust in a static light: as rigid structures that do not change over time. It is for this reason, we argue, that such accounts cannot explain instances in which ethnic institutions evolve and seem capable to adapt to the changing circumstances facing their members. In this Section we will discuss two widely cited contributions in this stream of analyses and we will point at problems related to their assumptions about the rigidity of the institutions they theorize.

The first is Avner Greif’s (1993) notion of a ‘traders’ coalition’, which he employs in order to analyse the business methods of the 11th century Maghribi traders: a group of Jewish merchants and their agents that conducted maritime trade in the western Mediterranean. According to Greif (1993, pp. 535-6), this group was initially formed and finally dissolved through political means: the expulsion of a group of Jewish traders from Baghdad in the first half of the 11th century and the imposed cessation of their trading activities by the rulers of Egypt in the end of the 12th.[2] It was a closed group -in the sense that membership remained unchanged for as long as the coalition remained in existence-which was embedded in the Jewish communities of the Western Mediterranean, yet with distinct cultural norms. Importantly, as we will see presently, membership was transferred from generation to generation.

Greif views this group as sustaining itself on the basis of strict reciprocity relations among members. If an agent cheated a merchant, he was sanctioned not just by the specific merchant but by the whole group, - i.e, the traders’ coalition pursued what Greif defined as a Multilateral Punishment Strategy (MPS) against cheaters. Thus, the coalition was the outcome of a long chain of repeated games, in which agents chose to cooperate rather than defect, as the payoff for cheating was perceived to be lower than the stream of payoffs they expected to reap from cooperation in an indefinite future. It is precisely these characteristics that explain the longevity of the coalition. This can be seen, by considering two related questions. The first is why an end-game was never played, i.e. why agents chose never to defect, even when in old age they should reasonably expect the stream of future payoffs to diminish. Greif’s answer is that the value of a trader’s reputation determined also his offspring’s standing in the coalition, thus discouraging opportunistic behaviour.

The second question is why a Maghribi trader might not choose to go into partnership with a non-Maghribi trader. Greif contends that in that case MPS would not be possible, as it rested on intense communication and information-sharing among the group’s members. But such communication channels between member and non-member communities did not exist, thus merchants preferred to cooperate with agents who were also members, rather than risk embezzlement in the absence of MPS. The manner in which Greif answers these two questions reveals the problem regarding its capacity to explain the longevity of the traders’ coalition;[3] both the assumption of unchanging membership –which was transferred from generation to generation- and the assumption of no business ties with non-members imply a rigid institution, which will remain in existence only as long as these characteristics remain unchanged.

The second contribution we address is the account of ethnic trading networks introduced by Jack Carr and Janet Landa (1994). Carr and Landa maintain that trust relations among the group’s members reduce the transaction costs associated with the risk of breach of contract by one’s contracting partner. Ethnic trading groups are quasi-intentionally created clubs[4] that aim at curbing opportunism. Thus they define (1994, p. 117) a club as: ‘any voluntary group deriving mutual benefit from the reduction of contract uncertainty… because: a) information about propensity to cheat can be obtained cheaper for club members than non-club members; and b) sanctions… exist and are imposed on members who violate the rules of the game set by the club, thereby making it more expensive to breach contracts with club members than contracts with non-club members’. [Our emphasis]. Interestingly, both characteristics of ethnic networks as clubs are remarkably similar to Greif’s ‘traders’ coalition’.

Carr and Landa treat ethnic clubs as voluntary –i.e. quasi-intentionally created- constructs. The implication is that a club will remain in existence for as long as members continue to believe that the probability of being cheated by a member is significantly lower than by a non-member. This is the transaction cost that, according to Carr and Landa, the formation of the club aims to curb. But given that the probability of breaching is all that differentiates transactions among members and non-members, isn’t it reasonable to expect that trust relationships will sooner or later begin to be forged between specific members and specific non-members? If that happens, the exclusive seeking of partners from within the club will begin to unravel jeopardizing the longevity of the institution.

3The problem of longevity: a simple model

In this Section we present a simple model that shows why this unraveling will inevitably be the case. We denote the traders supplying a market by Μ. Then there will be a subset, m, which comprises all elements of Μ that belong to an ethnic network. Therefore, all traders belong to one of two groups: Ai (i = 1, 2, … m), i.e. network members, or Aj (j = m+1, m+2, … M), i.e. the non-members. Every A agent can transact both with other members and with non-members.

We assume that the production of a unit of output requires the cooperation of two A agents.[5] We also assume that each A agent can go into partnership with only one other agent in every time period. The partnership expires in the end of the period and each agent is free to seek a new partner for the next. In this context, the analytically relevant question is whether a member of the network (Ai) will prefer to pair with another member (another Ai agent) or with a non-member (Aj).

We can conceptualize the problem as a game of musical chairs; all M agents run around a row of Q number of chairs, while music is playing. The moment the music stops some Ai and some Aj agents will find themselves seated, while some others will be standing. The ones sitting must now choose a partner to share their seats with. The question then is whether an Ai agent will choose to share his/her seat with another Ai or with an Aj agent.

This conceptualization of the problem helps to introduce and clarify some further features of the model. What motivates our agents is the pursuit of a profit opportunity –sitting in a chair- and the avoidance of staying out of business –finding themselves standing. We assume that each pair of agents can only produce a unit of homogeneous output q per time period –i.e. can only occupy one chair-, thus the scale of production is not an issue. In addition, the total output that can be produced in a time period (Q) –i.e. the number of chairs- falls in the following range:

m <2Q < M

This merely implies that production could not take place exclusively within the network, and that there is competition among agents to participate in partnerships. Finally, we assume that the business model is the same regardless of whether the pair of cooperating entrepreneurs consists of two Ai, or one Ai and one Aj agents.

In searching for a partner, every agent is motivated by the expectation of profit (EK), expressed simply as the difference between revenue (pq) and the sum of Production (PC) and Contracting Costs (CC). Thus, equation (1) describes the motivation of agents:

(1)EK = pq – PC – CC

However, this relationship can be refined further. Production Costs (PC) can be broken down to Input Costs (IC) and Organization Costs (OC). On the other hand, Contracting Costs (CC) can be broken down to Negotiation Costs (NC) and Breaching Costs (BC), i.e. the costs that an agent Ai will incur in the event that his/her contracting partner breaches the contract. Taking these refinements into consideration we have relationship (2):

(2)EK = pq – IC – OC – NC – BC

It is reasonable to think that Organization Costs (OC) and Negotiation Costs (NC) are lower when both partners are members of the network than when one of them is not. By contrast, we assume that Breaching Costs (BC) are identical regardless of whether the partners are members or not. However, following Carr and Landa (1994), we assume that the probabilities of Breaching Costs occurring are lower when both partners are members (πii) of the network than when they are not (πij), i.e. we generally assume that πiiπij. Thus relationship (2) can be expressed in the following two forms, depending on whether it refers to pairs of members or not:

If both partners are members of the network:

(3a)EKii = pq – IC – OCii – NCii – πiiBC

And if one partner is a network member and the other is not:

(3b)EKij = pq – IC – OCij –NCij – πijBC

Thus, an agent Ai will choose to partner with another member of the network (another agent of the m set) rather than a non member (an Aj agent) only if ΕΚii ΕΚij.

Under what circumstances will this hold? According to our assumptions, a number of terms in relationships (3a) and (3b) will be identical in the two partnering modes: we have assumed that each pair produces a homogeneous output q; the market price p will be the same for both modes, assuming perfect competition; finally, we have assumed that all partnerships use the same technology and business model, thus we can reasonably deduce that Input Costs IC will be the same in both modes. Therefore, for ΕΚii ΕΚij to hold unambiguously, the following must hold:

(4a) OCii OCij