What does it take for a roving bandit settle down? Theory and an illustrative history of the Visigoths

The present paper is motivated by McGuire and Olson’s perspective on state origination. Carneiro (1970) provides an overview of a broader swath of state origin theories.

Nozick (1974) provides a related description of state formation based on competition amongst private protection providers in an environment conducive to regional natural monopolies.

Rowley (2000) argues that in post-independence Africa the stationary autocrats recognized the tenuous nature of their regimes–that “they did not posses a durable monopoly in theft” (p. 139)—and therefore tended to act like roving bandits.

The same experiences have motivated other scholars to explore whether non-predatory, representative governments can arise directly from anarchy, foregoing any intermediate stage where a dominant warlord/stationary bandit arises (Wantchekon 2004).

DeLong and Shleifer (1993, p. 679) observe: “An absolutist government… comes close to Olson’s pure type of a stationary bandit or Brennan and Buchanan’s (1980) constitutionally unconstrained Leviathan.” To the extent that this is true—e.g., in the case of the French absolutism established in the seventeenth century—it remains irrelevant to issues of the roving-to-stationary transition unless one can point to empirical cases of absolutist roving bandits.

Furthermore, Caldara (2013) does not model the provision of public goods.

Relatedly, Young (2015a) studies much earlier societies of Germanic barbarians. He documents the institutional changes in governance institutions that occurred from the first century BC through the first century AD. A notable change was the transition from temporary military kings elected from the nobility to standing armed retinues led by entrepreneurial commanders. The armed retinues essentially were Olsonian roving bandits. The illustrative history provided in the present paper begins with fourth century Gothic retinues of this type.

Ertman (1997) emphasizes how the military revolution differentially affected the Latin European administrations of France and Spain versus that of Britain, with the former tending towards patrimonialism and the latter toward bureaucracy.

Herbst’s arguments are consistent with Kirby and Ward’s (1991) claim that the mechanisms underlying state formation in Africa are essentially the same as those described by Tilly. These authors argue that the key difference is lower productivity in Africa and therefore smaller returns to building state capacity (i.e., lower expected tax revenues).

Olson uses the example of a cartel working to establish an above-competitive price for its members’ output. Given the demand for that particular output, the more firms belonging to the cartel the less each member can sell and the lower is each member’s profits. Furthermore, if the cartel cannot restrict individual firms from over-consuming the above-competitive price (i.e., supplying too much output to the market for sale), then the exclusive good cannot be provided effectively.

Of course, one can imagine scenarios where a roving bandit expects to plunder a given out-group in the future and adjusts its current behavior accordingly. But this distinction between the games played by roving (non-repeated) versus stationary (repeated) bandits with out-groups is a.

Here I am assuming that anyone providing labor services for a stationary bandit is part of the in-group. Of course, once a stationary bandit recognizes an encompassing interest, the line between in-group and out-group blurs. (For example, considering the US government, who constitutes the in-group? Policymakers? Government workers? All of the taxpayers?) For now it will be useful to think in terms of a stationary bandit’s in-group as all agents acting under the bandit’s directives. For the literature on state capacity and economic development see Bessley and Persson (2009, 2010, 2011) and the additional references they provide.

A given supply of law and order can be subject to overcrowding/congestion, reducing the quality and quantity provided per capita. In this sense it is not a pure public good. See Comes and Sandler (1996) and Sandler (2013) for discussions of impure public goods.

The former must be greater than the latter, of course, if the in-group size increases overall.

In addition to the translation of Ammianus Marcellinus listed in the references, I have at times referenced the original Latin provided by http://​www.​thelatinlibrary.​com/​ammianus.​html.

Famously, Valens and Athanaric concluded a treaty while afloat on the Danube, an arrangement allegedly made so that Athanaric could obey his father’s command never to set foot upon Roman soil (Ammianus, Book 27, pp. 336–337). The standard translation involves Athanaric having sworn an oath to that effect. However, having been defeated by the Huns and his leadership abandoned by his Goths, Athanaric later was received by Theodosius I in Constantinople, where the erstwhile judge died. Wolfram (1975, p. 263) points out that the correct interpretation of the Latin is actually: “the Gothic chief was bound by an especially severe oath and that a paternal order had forbidden him to set foot on Roman soil.” If he did not break and oath, then, the passage begs the question of what exactly this severe oath was!.

As Ammianus (Book 31, pp. 416-417) dryly observes: “[T]he affair seemed matter for rejoicing rather than dread, and the practised flatterers in the emperor’s entourage extolled in exaggerated terms the good fortune which unexpectedly presented him with a large body of recruits drawn from the ends of the earth…. With these high hopes… the greatest care was taken to ensure that… none of those destined to overthrow the Roman Empire should be left behind.” Orosius (Book 7, p. 373) is more concise: the Goths were “received by Valens without negotiating any treaty. They did not even surrender their arms to the Romans.”

Orosius (Book 7, p. 380) counts this as a double victory for the Empire: “for the loss of these [the Gothic auxiliaries] was certainly a gain and their [the usurper and his followers] defeat a victory.” The leading Gothic men may have found the federate status more beneficial than the Gothic rank and file, a point to which I will return to in Sect. 4 below.

This was probably not true, but Stilicho had strongly supported granting the larger subsidies to the Visigoths in exchange for peaceful settlement in Illyricum.

This was much to the chagrin of Honorius who, “seeing that nothing could be done against the barbarians when so many usurpers were opposed to him, ordered that the usurpers themselves should be destroyed” (Orosius, Book 7, p. 393).

The question of why the Visigoths were settled in Aquitaine specifically is controversial. E. A. Thompson argues that it was to provide defense against rebellious Bacaudae peasants north of the Loire; J. M. Wallace-Hadrill (1967) argued that the Visigoths were placed there to counter Saxon and Frankish pirates in the Atlantic; and Vincent Burns argues that the settlement conveniently placed the Visigoths far away from the Mediterranean ports in Arles and Marseilles, with which the Romans were primarily concerned. See Burns (1992) for an overview of this debate and the relevant citations.

Fukuyama (2011, p. 74) refers to such armed retinues as “the most basic and enduring unit of political organization”; “such organizations became virtually universal in subsequent human history and continue to exist today in the form of warlords and their followers, militias, drug cartels, and street gangs.”

Club theory is associated with the seminal work of Buchanan (1965). Sandler and Tschirhart (1997) provide a review of the subsequent contributions. See Leeson (2011) on clubs with self-enforcing constitutions.

See Young (2015a) and the references therein.

Koyama (2012) provides an example from England in the 1800s. Private prosecution associations provided members with detection of crimes, the delivery of criminals to court, and the payment of court fees. These associations generated positive externalities to association non-members by way of deterrence effects.

Bó et al. (2015) discuss how, given initial conditions on productivity and the returns to investments, whether or not violence capabilities can be varied determines whether or not conflict tends to be exacerbated or moderated, allowing socially productive arrangements to emerge.

Emperors dealt with these small-scale raids in ways suggesting that they did not interpret them as orchestrated by some centralized organization. For example, Gothic raids into Thrace and Moesia typically did not provoke an overwhelming response from Rome. Anonymous Valesianus (ch. 5, p. 523) refers simply to Constantine’s “check of their attack”. And even during the period when Valentinian and Valens had granted federate status to some group of Goths, sporadic raids into Thrace and Moesia did not provoke a strong response. Alternatively, when a larger army of Goths (presumably a confederacy) attacked a group of Sarmatians with federate status in 323, Constantius’ (II; Constantine’s son and later co-emperor) winter campaign resulted in “almost a hundred thousand of the Goths […] destroyed by hunger and cold” (Anonymous Valesianus, ch. 6, p. 528). While this contemporary source (circa 390) may be exaggerating that number, there is no doubt that Constantius crushed a well-defined enemy. Similarly, when 3000 Goths joined in support of the usurper Procopious, Valens followed up with 3 years of campaigning across the Danube. This campaigning “reduced the barbarians to such want that they sent a number of delegations to beg for pardon and peace” (Ammianus, Book 27, p. 337). The reference to delegations also suggests that Valens was dealing with a Gothic confederacy. This conclusion also is supported by Ammianus’ (Book 27, p. 332) statement that the initial 3000 Goths supporting Procopious were sent by “their kings, now reconciled to Rome….” The plural “kings” again indicates that it was, again, a confederacy.

See Fukuyama (2011, pp. 56–59, pp. 72–75) for useful discussions of segmentary societies generally and the roles played by segmentary units in military organizations, including that of the Germanic barbarians recorded by Caesar and Tacitus. Baker and Bulte (2010) describe the agglomeration of smaller Viking groups. Kurrild-Klitgaard and Svendsen (2003, p. 261) provide further detail that is reminiscent of the Visigothic experience: “Raiding Vikings began to band together in large fleets and armies [and] sacked towns and markets…. Occasionally the forces disbanded, while at other times they re-united in the so-called ‘Great Army.’” Apparently small-scale roving Viking bandits also (at least initially) took part periodically in short-lived confederacies. Interestingly, the theory that Kurrild-Klitgaard and Svendsen (2003) provide to account for the Viking experience does not explicitly deal with the process of confederating.

This is consistent with the view of Wickham (2009, p. 45): “war encouraged the temporary development of alliances or confederacies of separate tiny tribes, each with its own permanent leader, but choosing a temporary leader for that confederation.” Contemporary Roman literary sources understandably emphasize the larger military conflicts. This has led some modern historians to emphasize the temporary confederacies relative to their constituent retinues. For example, Heather (1998, pp. 18, 57–58, 2006, p. 90) concludes: “By the fourth century… especially in areas close to the Roman frontier… smaller units had given way to fewer larger ones”. However, Heather (2006, p. 81) also notes that “in the fourth century, major conflicts also occurred only about once in a generation on Rome’s European frontiers”.

Given the treaty of 382 with Theodosius, the Visigothic confederacy became an entity with substantial autonomy though nested within the frontiers and institutions of the empire. Given that the confederacy was rooted in a general exodus of Goths across the Danube—including women, children, and the elderly—this likely gave what ultimately became the Visigothic leadership some experience in governance beyond that narrowly applicable to an armed retinue. This is also consistent with Thompson’s (1982, p. 38) comments: “The actual division of land in Moesia [where the Goths crossing the Danube initially were settled] must have been carried out in the main under the direction of the chiefs and councils, and we may be sure that the optimates [leading men; elites] thereby obtained executive power which would hardly have been theirs in the normal course of tribal life. Again, the annual subsidies of grain and cash paid over at times during the years 382–418 by the Roman government were delivered in the first instance into the hands of the chiefs and councils, who then proceeded to distribute them, or part of them, to the tribesmen in general.”

This is in reference to the fifth century (Visigothic) Code of Euric and the sixth century Burgundian Code.

The conventional view is that fifth and sixth century barbarian settlements occurred within the Roman legal framework for quartering soldiers (Gaupp 1844; Lot 1928). Each barbarian “guest” was allotted a fixed share of a Roman “host’s” property. More recently, Goffart (1980) argues that barbarian allotments were not actual land but rather units of tax assessment associated with the land. Sivan (1987) accepts that allotments of land were made, but argues that the legal framework was applicable to the settlement of retired soldiers rather than the quartering of active ones. See Young (2015b) for a discussion of the alternative interpretations of hospitalitas.

Olson (1993, p. 567) cited a quote from a monarchist in southern Italy (initially quoted in Banfield 1958) as motivating him to study the conditions under which both autocracies and democracies are conducive to economic growth: “Monarchy is the best kind of government because the King is then owner of the country.” See Polishchuk and Syunyaev (2015) for evidence on the importance of elites’ asset ownership for improving the provision of well-defined and enforced property rights. See Blaydes and Chaney (2013) for evidence from medieval Europe and the Muslim world supporting the complementary idea that such ownership aligns incentives more effectively when rulers’ tenures are longer. Also, Kurrild-Klitgaard and Svendsen (2003, p. 261 and p. 266) note that Vikings were often granted land by regions where they had or were likely to raid: “it gave the Vikings a long-term personal interest in defending the area and preventing others from attacking it.”

Some scholars argue that Alaric II was the promulgator of the Code of Euric as well as the Breviary (Heather 1998, p. 184). Wickham (2009, p. 88) observes that “Visigothic… kings legislated, taxed, shipped grain around, used Roman civilian officials, and created integrated Roman and ‘barbarian’ armies, including Roman generals.” While I focus on the incorporation of aristocratic human capital here, Wickham’s observation indicates that the Visigoths were as well also open to incorporating other elements of Roman society into their in-group. Heather (1998, p. 193) echoes Wickham’s observation on its last point: “From c.470, for instance, we find Romans holding military offices to which they had been appointed by [Visi]Gothic kings….”.

This fits the case, described by Leeson (2007b, p. 306), of “strategies employed by permanently weaker individuals to alter the incentive of stronger agents for trade versus banditry.” By becoming essential components of the barbarian bureaucracy and courts, the Gallo-Roman aristocrats were able to trade their services with, rather than just be the prey of, the Visigoths. Relatedly, one notes that the Gallo-Roman aristocrats came also to dominate ecclesiastical positions. The Church provided an institutional check on the government of the (Christian) Visigoths (Mathisen 1993).

Halsall (1992, 2007) argues that the withdrawal of imperial military presence in the north began as early as the mid-fourth century. Regardless of the precise dating, sometime at the end of the fourth century the seat of Gaul’s praetorian prefect moved from Trier to Arles (Mathisen 1993, pp. 19 and 59).

“Under Gundobad’s [ruled 474–516; son of Gundioc] leadership the Burgundian kingdom reached its greatest extent: the northwest it extended as far as Langres; the northeast to the northern Iura Mountains; to the east to the Alps; to the west it was bounded by a portion of the Rhone River and the upper course of the Loire; and for a time the Burgundian kingdom also included Provence to the south, but this was soon lost to the Ostrogoths” (Drew 1976 [1949], p. 2).

For examples, the Visigothic Code of Euric states that “Ancient boundaries are to stand, just as our father of blessed memory prescribed in another law”; and a Burgundian law states that if a Burgundian wishes to sell his land, his Roman “host” has to be given preference before contracting with anyone else.

For basic facts regarding Theoderic the Ostrogoth, see Hodgkin (2011 [1891]).

While the Ostrogothic Theoderic achieved historical designation as the Great, the defeated Gothic confederacy coincidentally was led by an individual known to us as Theoderic Strabo (the squinty-eyed).

When Theoderic defeated Odovacer’s forces, the Ostrogoths still faced a long and costly siege of Ravenna where Odovacer was holed up. Theoderic ended the standstill by signing a treaty whereby both he and Odovacer would rule Italy jointly. However, at a banquet organized by Theoderic to celebrate the treaty, Theoderic made a toast and then drew his sword and murdered Odovacer with a brutal strike from above and down through the latter’s collarbone.

In the case of the Franks, this situation may have changed following Clovis’ short-lived consolidation of the Frankish kingdoms. See below.

Britain (and also, to a lesser extent, northern Gaul) suffered from substantial disruption of the trade networks linking it to the Mediterranean economy (Rauch and Michaels 2013). Complementary to the arguments made here, if less wealth simply was available to tax in the north then this would have offset the tendency towards barbarian state formation. However, whether or not Britain ultimately was substantially poorer than the rest of the Western Empire is controversial. See Heather (2009, pp. 268–274) for a discussion of the controversy and the contrary view that “Roman Britain was no backwater” (p. 269).

The distinctive history of regional government in England—shires (counties) and hundreds—is a direct legacy of the Anglo-Saxon migration/invasion and the subsequent proliferation of numerous small Anglo-Saxon polities.

In 1653, Childeric’s tomb was discovered near Tornai. The Merovingian’s identity was confirmed by a signet ring inscribed with his name. Among other items, the tomb included about 300 small golden bees. Napoleon would later harken back to the Childeric and the Merovingians by replacing the Bourbon fleur-de-lys with a bee as the symbol of imperial France (James 1988, p. 61).

The disintegration of Clovis’ kingdom was surely explained partly by Frankish customs regarding the division of property across all male heirs (James 1988, p. 91). That being said, the disintegration is indicative of the fact that a Merovingian state separate from the personality of its king failed to emerge. Also, one might note that the Ostrogothic Kingdom likewise disintegrated shortly after the demise of Theoderic. However, in that case Theoderic failed to produce a male heir and, furthermore, the Ostrogoths had to deal with the eastern emperor Justinian’s efforts to reclaim the territories of the Western Empire.

Despite the fact that Clovis’ Franks had by 508 driven the Visigoths out of Gaul and into the Iberian Peninsula, the Visigothic Kingdom (regardless of its precise borders) was a more durable state than ever formed in Merovingian Gaul. Theoderic I established a Visigothic dynasty in 419 that would last for more than a century and the Visigothic Kingdom lasted through the seventh century until the Muslim conquest. Alternatively, Clovis’ sons (and their successors) presided over small kingdoms that had turbulent relations with one another, often in conflict—sometimes violent—with one another. See Gregory of Tours’ (1974) History of the Franks Book III, ch. 1.

The out-group can also have within-group interactions that reduce the incentives to produce (e.g., out-group members constantly may be stealing from one another.) The out-group might also be more productive and prefer to live with rather than without a stationary bandit if the latter provides internal law and order to the former.