2. The necessity of a historical approach

Hicksian models are typically motivated not by the historical emergence of monetary exchange, but by current monetary policy concerns. Two different frictions that both result in the emergence of indirect exchange in a GE economy may have very different implications for its reaction to a monetary shock, for example. The financialized market economy to which they are intended to apply is characterized by strong and distinct equilibrating forces, which justifies a solution with a unique symmetric equilibrium (cf. Koppl Reference Koppl2002: 94). The initial conditions of these models do not correspond in any meaningful way to the conditions out of which monetary exchange in fact emerged, nor are they meant to. For this reason they should be interpreted as fundamentally synchronic rather than diachronic, despite their featuring ‘emergence’ of money as an equilibrium result.

The equilibrating forces characteristic of a monetary economy cannot be taken for granted in a model of the development of monetary exchange. The increasing returns model in this paper has no such neat result as the emergence of a single medium of exchange out of the interaction of optimal trading strategies. Instead, it advances a broad framework through which the selective forces propelling the development of exchange institutions can be understood and organized.Footnote ² These institutions are the crucial missing piece that renders Hicksian models inapplicable to the actual historical emergence of monetary exchange.

An exchange institution is a set of conventional practices that facilitates the incentive-compatible exchange of goods and services. Monetary exchange in particular is an institution where exchanges are facilitated by the common acceptance of one or several media of exchange – money – in terms of which the prices of other goods are quoted and rendered comparable with one another.

The necessity of exchange institutions is often overlooked, both in economics broadly and monetary economics more specifically. John Stuart Mill (Reference Mill1848, bk. 3, ch. 7, §3), anticipating Hicks, argued that ‘there cannot, in short, be intrinsically a more insignificant thing, in the economy of society, than money . . . It is a machine for doing quickly and commodiously, what would be done, though less quickly and commodiously, without it.’ Abstraction from the particulars of exchange institutions has been standard practice in economics ever since. Indeed, because exchange benefits both parties, it may seem not to require any particular explanation. Adam Smith (Reference Smith1776, bk. 1, ch. 2, §1) referred to the ‘propensity to truck, barter, and exchange’ as a basic feature of human nature, and most economists have taken this assumption as axiomatic.

Nevertheless, the ability to truck, barter and exchange runs into a number of impediments that are typically assumed away in Hicksian models of monetary exchange, where the institutional background is more or less invisible. In particular, non-trivial institutions must be present in order to assume either (1) pre-existing specialization, or (2) straightforward property arrangements.

3. A prehistory of the division of labour and the extent of the market

In light of these restrictions on our initial conditions, we begin with autarky – the size of the exchange network being zero, and therefore without specialization. This will give us a sense of the basic impediments to the regularization of exchange that any exchange institution must solve.

Consider the prospects of two identical and autarkic foragers, Crusoe and Friday, arranging an exchange. We must suppose the first exchanges to happen under conditions of zero endowment: the existence of individualized endowments presumes a pre-existing property norm (i.e. uncontested control of goods), and the exchange of differentiated endowments further presumes pre-existing specialization. For this reason, our prototype must be the reciprocal exchange of undifferentiated services. More specifically, services performed on the body of the other party – for example grooming, which must be performed sequentially – are a more plausible prototype for the first exchanges than other services that might be performed simultaneously, because the former are easier to monitor.

Crusoe and Friday are, by hypothesis, both better off exchanging grooming. But both would also rather receive a grooming and then go swimming without returning the favour. By the same token, conditional on not receiving a grooming, each party would rather avoid the hassle and stay at the ocean. Despite the potential gains from trade, the equilibrium outcome is no exchange. The basic problem with getting trade started in the first place is that the time-separated quality of service exchange poses a prisoner's dilemma and opens the door for defection.Footnote ⁵ Without this first step, there is no opportunity for the development of specialization and the exchange of goods.

The primary impediment to exchange at this point is not search costs, since at this stage any individual is more or less as good as any other at performing an available service, but a commitment problem. The pertinence of the dilemma can be seen even more readily in the animal kingdom, where the fitness costs of being suckered preclude most forms of cooperation and exchange.Footnote ⁶ Even in this setting completely lacking in specialization, if Friday cannot commit to return the grooming, there is no exchange.

Dilemmas such as that faced by Crusoe and Friday are pervasive impediments to exchange and social organization (cf. Miller Reference Miller1992). The distinctiveness of human social behaviour is not, therefore, as is usually assumed, merely that man is able to appreciate gains from trade whereas animals are too dull to do so. Rather, humans are distinctive in their ability to commit on a non-rational basis to act contrary to their own narrow interests, though these departures from narrow-sense rationality are themselves ‘rational’ in a broader sense (Frank Reference Frank1987). It is not because humans are more rational than animals, but indeed because they are less narrowly rational that they can be trusted, on the whole, to reciprocate cooperative behaviour even at some cost to themselves.

4. A taxonomy of exchange institutions

With autarky as the backdrop, the problem facing any exchange institution will be: how to overcome the prisoner's dilemma of reciprocal altruism and make exchanges self-enforcing? There are factors that both help and hinder as we move toward a consideration of more familiar large-scale human interaction. Repeated play allows Crusoe to punish Friday if he defects or free-rides. However, this equilibrium becomes more and more difficult to enforce as Friday's brothers, Wednesday and Thursday, arrive on the island. As Bowles and Gintis (Reference Bowles, Gintis, Blume and Durlauf2008) show, repeated play on its own is generally insufficient to sustain cooperation in groups larger than about four.

Cooperative exchange in a larger group therefore requires an exchange institution, which consists of norms and strategies that facilitate:

1. 1. The accounting and publication of exchange histories to identify defectors at low cost, and

2. 2. Convergence upon a coordinated strategy to punish defectors.

The folk theorem implies that the space of potential exchange institutions is extremely vast. Nevertheless, because the costliness of complex strategies will tend to select against them if they fail to provide a corresponding benefit, we can imagine a rough correlation between institutional complexity and the size of the exchange network. This will allow us to categorize exchange institutions along a single dimension of strategic complexity.Footnote ⁷

The basic problem with increasing the size of the exchange network is that the accounting necessary to prevent defection from becoming a dominant strategy as group size increases requires increasing strategic complexity, which in turn requires increasing cognitive capacity (Dunbar Reference Dunbar1992, Reference Dunbar1995). Humans do have an advantage here among animals: from their very emergence as a separate species, anatomically modern humans have been characterized by efficient facial recognition (Haxby et al. Reference Haxby, Hoffman and Gobbini2002) and a relatively large memory for keeping track of personal obligations. Still, impressive as it is compared to other primates, humans’ natural capacity for accounting obligations allows them to maintain group sizes averaging a few dozen, and maxing out around a few hundred.

However, humans do have another advantage: the ability to extend their cognition by imbuing elements in their environment with symbolic significance (Clark and Chalmers Reference Clark and Chalmers1998). Writing is the prototypical example: we can extend our effective memory by replacing huge quantities of information with a reference, i.e. with the knowledge of where to find it. Similarly, investment in the capacity to offload the accounting of exchange balances into the environment has the potential to vastly increase the information taken into account when making an exchange, and therefore also the scale and complexity of social organization.

We can use this idea of extended cognition to operationalize our taxonomy in terms of fixed and marginal costs. The fixed costs of an exchange institution are the opportunity costs for a society of converging onto and socializing its members into certain norms, symbols and meanings necessary for extending cognition, taking biological and evolutionary costs as given.Footnote ⁸ To put it another way, any pattern of behaviour can in principle be identified as having a joint basis in biology and culture. The cultural pattern may build upon, interact with, or even override biological patterns (cf. Hayek Reference Hayek1960: 93). The learning and habituation necessary to maintain an institution – i.e. its fixed cost – will be higher the greater (1) its cultural as opposed to biological basis, (2) the extent to which it conflicts with or overrides biological predilections, and (3) the information set upon which it conditions action (i.e. its complexity).

The marginal cost, on the other hand, is the cost of executing a particular exchange, including the risk of defection by the other party. Because costly but ‘functionless’ practices will tend to be selected against in the long run, marginal costs will be a diminishing function of an institution's fixed costs. On the accounting side, the more extensive the convergence on accounting strategies and symbols, the less cognitively costly exchange will tend to be on the margin. And on the punishment side, the more effective the punishment strategy, the lower the risk of defection. In both cases, lower marginal costs increase the potential scale of human organization.

The investment in symbolic convergence that constitutes an exchange institution enables an increase in strategic complexity without a corresponding increase in on-board cognitive capacity. To anticipate the following sections somewhat, the organization of exchange using only the natural cognitive faculties (what we will call ‘tribal credit’) involves a high marginal cost for each relationship – i.e. some relatively high portion of the brain's social memory. Despite its low fixed cost, the high marginal costs of maintaining ongoing relationships mean that tribal credit exhausts cognitive limits at a very low scale. Monetary exchange, by contrast, involves high fixed costs but minimal marginal costs. Any given transaction requires relatively little thought to accomplish, and once completed, can be safely forgotten in a way that would destroy the self-enforcing quality of tribal credit exchange.

The formulation of exchange institutions in terms of fixed and marginal costs suggests an analogy to capital theory. Lachmann (Reference Lachmann1956: 80) asks the same question of physical capital that we have asked of institutions:

Similarly, even for a society on the brink of subsistence, investment in high-fixed-cost institutions is not an equilibrium outcome until a sufficient volume of exchange has built up to amortize the costs. Tribal foraging societies, limited as they are in scale, simply cannot sustain enough exchange to make the fixed cost of monetary exchange – namely, a system of writing and mathematics, along with appropriately credible institutions – worthwhile to develop and sustain.

This process of mutual feedback brings us back to the bootstrap problem (cf. Kaldor Reference Kaldor1972): if specialization and the volume of exchange are mutually limiting, and if zero-fixed-cost institutions are a stable equilibrium, how does the process of investment in higher-fixed-cost exchange institutions ever get off the ground?

Because an instrumental approach to the development of an institution would tend to destroy its commitment power, we must suppose the process to be driven without anyone deliberately aiming at the result. A process of variation and selection fits the bill. If we assume some degree of random variation in exchange institutions, population pressure will select for higher-fixed-cost institutions, simply because a mismatch where the size of the society significantly outpaces the exchange institution will imperil that society's continued survival. A society with numerous groups in close proximity must find a way to live peaceably with most of them most of the time, or face perpetual conflict and possibly extinction. In this sense, whatever other aspects it may have, any integrative institution that results in expanding the feasible scale of social organization must necessarily be an exchange institution – i.e. it must involve strategies for accounting and/or punishment.

One implication is that any ‘big push’ effort to monetize a society is likely to fail, for the same reason that Lachmann's theory predicts the failure of ‘big push’ investment as a development strategy. Both elements of the mutual feedback must proceed roughly pari passu. If social scale outpaces the supporting exchange institutions, it must not be to such an extent that the discovery of and acclimation to new institutions is out of reach, in which case a defection cascade and social collapse will ensue.Footnote ⁹ By the same token, higher-fixed-cost exchange institutions enable increases in social scale but do not drive it. If institutional investment outpaces social scale, the result is likely to be wasted effort, and eventual atrophy of a functionless institution.

As evidence for this account, a striking feature of contemporary tribal credit societies is that they persist in areas with little to no space constraint.Footnote ¹⁰ Environmental pressure, then – whether high birth rates or more intense competition for land – is a likely candidate at least for the initial shock that would push a society out of the zero-fixed-cost equilibrium, and into either extinction or a higher-fixed-cost exchange institution.

5. The tribal credit economy

This zero-fixed-cost case, relying entirely on the brain's innate capacities for accounting and publication, will correspond to the earliest known form of social organization in cognitively modern humans, as well as the form characterizing the most primitive hunter–gatherer and subsistence agricultural societies today. We will call this the ‘tribal credit’ economy.

Tribal credit is the terminus of a more or less straightforward progression from autarky through dyadic and small-group cooperation as cognitive capacity increases. Most importantly, the time-separated quality of exchange is preserved through the whole chain, even as the exchange of fishing for gathering becomes reified into an exchange of fish for berries – hence the ‘credit’ aspect of early exchange (Wray Reference Wray2004), ruled out by hypothesis in Hicksian models. As Wiessner (Reference Wiessner1977) notes, the term ‘gift economy’ obscures this exchange aspect: despite the nominally gratuitous transfer of goods, participants keep mental accounts of who owes how much to whom, and ‘woe to him who does not [eventually] make a corresponding present in exchange’ (Simmel [1907] Reference Simmel1978: 95).Footnote ¹¹

Tribal credit is named after its key institution, the tribe, within which both accounting and punishment are performed in a relatively decentralized manner. Regular interaction with a more or less definite group of people facilitates the regular dissemination of reputational information. Punishment, likewise – both of violence and of free-riding – consists of diffuse social pressure that can escalate into ostracism or death. Each member has familiar knowledge of all or most other members,Footnote ¹² and the ‘trading history’ of each member – unlike under atomistic barter – is common knowledge.

With the regular publication of reputations and the collective exercise of social pressure on defectors, the tribe is able to harness the time-separated quality of early exchange to create mutual rents among tribe members sufficient to enable them to commit to cooperate with one another for the foreseeable future. In contrast, simultaneous exchange – by giving up the leverage of repeated play within a local structure – would make it relatively infeasible to punish defection.

Unfortunately, the marginal cost of exchange under tribal credit is extremely high, which places a hard limit on both group size and the division of labour. Because cooperation in a reputational game depends on knowledge of the identity and history of each potential trading partner, it breaks down when enough players are anonymous. The scale of tribal credit organization is therefore limited by humans’ cognitive capacity for keeping track of relationships and obligations, thought to average around 150 relationships (Dunbar Reference Dunbar1992, Reference Dunbar1995). Indeed, in the absence of investment in higher-fixed-cost integrative institutions, tribal credit societies cannot get far past 150 members before free-riding and conflict become endemic. Such societies invariably have processes for fissioning the tribe or village after reaching a few hundred people (Bandy Reference Bandy2004; Chagnon [1968] Reference Chagnon2009). As a community approaches the size where any two individuals are as likely to know each other as not, free-riding becomes the dominant strategy, conflict rends it in two, and one faction is forced to establish itself elsewhere.

With the extent of the exchange network thus limited, the division of labour under tribal credit remains mostly ad hoc (cf. Hooper et al. Reference Hooper, Demps, Gurven, Gerney and Kaplan2015), with little capacity for permanent specialization beyond gender roles. There are simply not enough potential exchange partners for it to be worth it for any of them to relinquish food production and to specialize in, say, pottery manufacture.

Such societies must form the starting point for a theory of monetary exchange. Unlike Hicksian models of atomistic barter, where commitment is impossible and unlimited cognitive power allows for costless optimization, early human society is characterized by robust institutions for long-term commitment and extremely limited accounting capacity. Though there are no markets per se in a tribal credit society, even a basic organization of in-kind multilateral exchange offers sufficient incentive for defection that mechanisms for accounting and punishment are necessary to make it work. Tribal institutions, by making exchange ongoing and personal, enable sufficient commitment to sustain cooperation at a scale of a few dozen to a few hundred individuals. As compared with Menger's story, and without permanent specialization, these credit relationships render in-kind and time-separated barter sufficiently tolerable to forestall the spontaneous development of indirect exchange.

6. Trading networks

Exchange at a supra-tribal scale will require institutions to facilitate exchange without the extensive personal information that the tribe provides. The first step, historically, was the formation of inter-tribal trade networks, which allowed trade to take place on the basis of group affiliation rather than personal knowledge.Footnote ¹³

There are two basic preconditions for the formation of trade networks. First, agents must be organized into groups within which some combination of internal and external forces suppresses defection. Leaving external enforcement to the side until the following section, our question will be: how can a community organize itself internally in order to commit credibly to forswear opportunism vis-à-vis other communities, and what forces lead it to do so? This is precisely the function of tribal institutions from the perspective of an outsider: by suppressing internal defection, they allow the group to act as a ‘superorganism’ (Wilson and Gowdy Reference Wilson and Gowdy2015) and ensure its trustworthiness vis-à-vis outside groups.

Second, agents must be able to use information about the reliability of the defection-suppressing mechanisms of another class or tribe as a substitute for knowledge of its individual members. This allows agents to respond to trustworthiness with trust, and in turn makes trustworthiness a worthwhile investment. Marwick (Reference Marwick2003) argues,

To the extent that group affiliation can substitute for personal knowledge in the evaluation of trustworthiness, higher-fixed-cost institutions can regularize ad hoc trading into permanent inter-group specialization, with the trading relationships between groups analogous to the relationships between individuals in the previous section.

Inter-group trade is the prototype of simultaneous exchange. Indeed, demanding repayment at the time of exchange is seen in tribal credit societies as a peculiar form of trade reserved for ‘outsiders’, and protects to some degree against defection in a situation where the two groups are not in regular enough contact to make something like tribal credit relationships feasible on an inter-group scale.Footnote ¹⁴ A small number of outside trading partners makes simultaneous inter-group exchange far more manageable than it would be in the intra-group network of dozens or hundreds of people.

Consider a society consisting of an inland and a coastal village. Because their ecologies are so distinct, the gains from trade are readily apparent: it pays to organize an ongoing division of labour between the coastal village, which provides fish, and the inland village, which provides vegetables. But even with simultaneous exchange, the two villages still face the prospect of defection. Specifically, permanent specialization implies reliance on the other party, which allows it to ‘exploit’ the first for better terms of trade (cf. Klein et al. Reference Klein, Crawford and Alchian1978). In order to make specialization worth the cost, the village must be assured of a steady (or at least a predictable) demand for its product. Even if the gains from trade are substantial, a society just embarking on the division of labour will have limited means of smoothing consumption. For a risk-averse agent, an increase in the variance of their consumption is only worth it for a substantially higher mean (see the discussion in Wiessner Reference Wiessner1977).

For this reason, inter-tribal trade tends to be richly laden with ceremonial significance. Ceremony and religion are examples par excellence of fixed costs of exchange institutions, in the sense of imbuing certain objects or actions with additional symbolic significance for the purposes of coordination. The ceremonial organization of exchange facilitates coordination and goodwill by sanctifying certain arrangements and removing them from the set of negotiable terms (Leeson and Suarez Reference Leeson and Suarez2015). In other words, restricting the scope of deliberation can convert a prisoner's dilemma – i.e. the negotiation of terms afresh with every interaction – into a coordination game – i.e. convergence upon the significance of the relevant ceremonial symbols (Bear et al. Reference Bear, Kagan and Rand2017). Indeed, the more permanent and integral a particular trading relationship, the more elaborate the ceremony is likely to be (see e.g. Malinowski Reference Malinowski1926, ch. 3).

7. Managing specialization

In a fairly small or sparsely populated society, the probability of failing to find trading partners willing to buy one's wares in sufficient quantities to safeguard against starvation is high enough that investment in most specialized human capital is not worth the risk. As the size of the trading network increases with the development of inter-group trade, groups are able to settle into permanent patterns of specialization and individuals are able to find productive ways to specialize in intra-group roles. Increased demand for particular products arises with specialization in production.

This section, then, describes investment in two different directions to manage incipient specialization – namely, indirect exchange, which can develop in a quasi-tribal structure, and customary accounting, which is largely a political innovation upon which economic structures can build. The later convergence of the two, with indirect exchange augmented by precise accounting, we will call monetary exchange.

8. Market prices and the mass money economy

Monetary exchange is not a sufficient condition for the economic development we enjoy in the modern world. Customary exchange shades into monetary exchange over the course of vast expanses of time; indeed, several millennia elapse between the development of cuneiform writing in ancient Sumer and the first coinage in 7^th-century BC Lydia. And from the beginning of coinage, more than two millennia elapse before Europe's ‘great divergence’.Footnote ¹⁹ Something of the character of monetary exchange changed over the course of the 18^th and 19^th centuries, beginning in England and radiating out to Western Europe.

The bootstrap problem points toward an explanation. The problem is not merely to get the mutual feedback process going, but to keep it going, rather than settling into a mutually reinforcing equilibrium at a particular level of investment. While selection due to population pressure imparts some motion to the system, the major shift characterizing the era of economic growth – what we will call the mass money economy – appears to be that, at a threshold level of specialization, technology begins to propel the feedback loop: increases in specialization make it possible to develop and profitably invest in market-expanding technologies that further increase the scope for specialization, and select for higher-investment exchange institutions much more quickly than had been possible to that point.

9. Institutional disinvestment

In order to avoid the impression of a Whig theory of the history of exchange, the possibility of disinvestment in exchange institutions must be acknowledged as well. There are numerous historical examples of such disinvestment following a collapse in the size of the exchange network. Humphrey (Reference Humphrey1985), for example, documents ‘post-monetary’ societies whose use of indirect exchange atrophies after the severing of an important trade route.

The most significant Western example is the increasing unreliability of the monetary unit at the fall of the Roman Empire, and later the Carolingian empire. The collapse in both international and domestic trade triggered a dramatic ruralization and spontaneous dioecism throughout Europe, and Rome's money economy to fall into disuse (Kohn Reference Kohn1999). Bloch's description of the rise of feudalism (Reference Bloch and Postan1966: 250f) can be understood in these terms:

In other words, at some threshold market size between that of Roman cities and rural estates, and without a reliable monetary unit, specialization became a less attractive prospect. Enough people returned to subsistence agriculture that the whole process began to unravel, and Europe fell back for a time to an early customary exchange equilibrium. Without specialization, there was little need for a medium of indirect exchange, without which there was little need for numeracy and accounting skills. As these began to atrophy, the monetary economy became more and more defunct until feudal organization was settled into as ‘the line of least resistance’ – simple enough in its basic form to be conducted largely without the aid of arithmetic or money. It was not until the 12th century, when long-distance trade again began to increase the size of the market, that Europe was able to remonetize, and even then only in fits until the 17th century (Spufford Reference Spufford1988).

One must hope that nothing so cataclysmic awaits the developed world today. Nevertheless, the slowdown in global trade since 2010 (Hoekman Reference Hoekman and Hoekman2015) and the ballooning of unfunded obligations throughout the developed world over the past half-century have the potential to trigger the ‘liquidation’ of prior investment in exchange institutions, and especially in those more recent institutional developments that credibly restrain political actors from manipulating their currencies too extensively in their own favour through debt monetization. Such a breakdown might necessitate, if not a return to customary exchange, at least a return to full-bodied currencies as a prophylactic against overly extensive manipulation.