The Trouble With Guilds

For a rent-seeking organization that excluded competitors, fleeced customers, and underpaid suppliers, the European guilds have been staunchly defended by economists and economic historians—so much so that a pro-guild “orthodoxy” had emerged by the mid-2000s. The debate over whether these institutions were actually efficient solutions to market failures is a fun one, but alas, the evidence remains mostly anti-guild.

---

In researching for my recent post on “The Mechanics of the Industrial Revolution,” I noticed that a key explanation for the purportedly higher levels of artisanal skills in Britain relative to the Continent was the weakness of the former’s guilds. “[T]he relative limited [sic] power of guilds meant that rapidly growing sectors could swiftly attract extra apprentices,” who in turn served as the skilled labor force manning British textile machines and ironworks. Guilds had resurfaced in European cities after the six hundred-year hiatus of the Dark Ages after 1000 as trade and manufacturing began to concentrate in urban areas. They allowed artisans and merchants to organize and bargain for favors from local potentates, usually exclusive rights to sell goods and to regulate entry into the practice; the monopoly rents so derived were in turn shared with the elites who legitimated the system.

None of this sounds particularly good for growth. But for several decades up to the mid-2000s, economic historians developed a litany of arguments stressing the efficiency of guilds—that is, that these institutions created aggregate economic benefits that outweighed the obvious costs. There are four-ish main planks of what Sheilagh Ogilvie terms the “guild enthusiast” argument:

1. Guilds upheld quality standards and resolve information asymmetries.

2. Guilds transmitted skills and training.

3. Guilds facilitated technological innovation.

4. Guilds improved urban politics.

The major debate on the issue came in the pages of the Economic History Review during 2007-8, as S. R. Epstein and Sheilagh Ogilvie disputed all of these main points. Epstein unfortunately died before he could respond to Ogilvies’s rebuttal, but the exchange is worth understanding nonetheless. My description of the altercation is based largely on Epstein’s 2008 paper, Ogilvie’s reply, and her contemporaneous working paper that expands on the critique.

Quality Control

To Epstein, quality control—ensuring high product standards—was one of the central purposes of a guild. The primary evidence that guilds actually provided quality control comes from guild legislation, which frequently did seek to fix quality standards. But as many historians point out, written records can be a poor guide to actual practice. Ogilvie conducted a case study of the strongly-regulated Württemberg worsted industry to provide empirical evidence for these theoretical concerns. The actual punishments levied against offenders tended to be so trivial (apologies, minor fines, promises of reformation) as to be inadequate to keep members’ standards high, and—when enforced–were frequently meted out largely to keep cheaper competitors off the market. Price and wage ceilings actually harmed quality. Furthermore, quality offenses were so frequent that it seems dubious that the restrictions were particularly oppressive for would-be violators. Patrick Wallis, for example, found that in the early 17th century, 30 percent of the London Apothecaries and Stationers’ Companies had broken the rules.

Epstein objected to most of Ogilvie’s case study. On quality, he asserted that the examples given of guild regulatory inefficacy were unrepresentative—a claim that appears to be patently false, given that examples from twelve different worsted industries are given—and that Württemberg was doomed to be poorly enforced anyway because it was a proto-industrial area with dispersed production. This latter point also appears to be incorrect, and in any event, a common argument for the necessity of guild controls is precisely the fact that information asymmetries are rampant in dispersed putting-out systems. The Württemberg study ultimately found guilded products to be objectively poor, and that they also prevented sellers from marketing lower-quality products to consumers who actually wanted them. All of the above imposed deadweight losses, as large volumes of potentially beneficial transactions couldn’t take place.

Training and Skills

Another key argument for the economic utility of the guilds is that—through their apprenticeship system—they provided training for entrants into skilled crafts that could not otherwise have been acquired. In many areas of production, this just wasn’t true. The worsted textile industries studied by Ogilvie appear to have been open (sans guild restriction) to new, “untrained” entrants—indeed, hence the guilds’ incessant war on illegitimate producers. For example, she cites a 1582 petition by an urban guild of woolen-broadcloth-weavers against competitors who were seizing export markets “after learning combing and weaving for only a few weeks or months.” If the untrained were incapable of participating in a trade, then why were the guild masters so worried about them trying? In trades where guild apprenticeship was not mandatory, it was deliberately circumvented by people who realized how little they needed to use it. In the Yorkshire worsted industry, an increasing division of labor and the absence of such regulations meant that apprenticeship became increasingly outmoded and ignored.

If guilds were essential for the transmission of skills to the broader workforce, then the fact that they restricted entry into trades was probably counterproductive. Epstein had originally argued that entry was not rationed in Ogilvie’s German case because masters could take unlimited numbers of apprentices, but the worsted guilds had actually imposed quotas in 1611 and ratcheted them upward in subsequent years. Though the number of masters did rise, this was actually a result of internal increase (masters’ sons entering).

Some of Epstein’s arguments are just weird. He at one point completely misquotes Ogilvie, claiming that she wrote that ‘labour productivity “should have been higher in the countryside, where corporative regulation might be expected to be weaker,”’ when she actually wrote that “[s]tandard assumptions about proto-industrialization would predict higher output quotas in the countryside, where corporative regulation might be expected to be weaker.” He thus attributes to her an argument that she is actually attempting to debunk and uses her evidence on output quotas to show that countryside labor productivity was higher when she wasn’t talking about it at all!

Epstein also asserted that the fact that apprentice fees were paid and training years endured evinces the utility of the program itself, when it’s pretty obvious that trainees would be perfectly willing to waste time given that it was the only way to get certification as a master. Apprenticeship fees were not bonds to ensure future performance, either, because at 14-24, payers were already expected to be at something close to adult productivity, and thus able to reward masters financially.

Also, is this the best paragraph in the history of the EHR?

Epstein’s criticism of my German evidence is again vitiated by errors and distortions. His inclusion of masters’ daughters among women who could do craftwork freely is simply false. His claim that spinners demonstrate that nonguild-related females could work at guilded crafts omits to mention that guilds restricted them to spinning, prosecuted all other female ‘encroachers’, and capped spinners’ wages. His accusation that I assume female labour productivity to equal male is false: I explicitly analyse gendered productivity differences. His implication that guilds’ exclusion of females is justified by women’s domestic responsibilities and low productivity is called into question by demographic realities and guild masters’ opposition to women as skilled competitors. Epstein’s reinterpretation of woollen-weavers’ objections to short worsted training as a defence of their guild’s reputation is unjustified, since worsted wares were new in 1582 and woollen-weavers operated in different markets. Epstein’s assertion that nonguilded labour could not compete with guild-masters’ ‘all-round expertise’ is unsupported by the Yorkshire study he cites and is undermined by masters’ opposition to unapprenticed workers as dangerous competitors. His claim that when apprentices absconded they were manifesting a ‘sophisticated understanding’ of the importance of guild training is far-fetched beside the alternative interpretation that apprentices who quit had decided guild training was useless. Epstein’s statement that Württemberg worsted guilds did not restrict apprentice numbers is simply false. His notion that outside entry declined after 1650 because no one desired to enter is refuted by voluminous evidence of outside applicants rejected by the guild. His claim that the guild cannot have become more exclusive after 1650 since numbers increased ignores internal recruitment of masters’ sons. His assertion that average output fell after 1650 because of weak demand for guild jobs is based on careless misreading of evidence on guild output quotas. His argument that apprentices would not have paid guild fees had they not valued guild training is ridiculous: becoming a journeyman or master required an apprenticeship certificate, issued only if fees were paid. Epstein’s claim that my evidence shows labour productivity was higher in guilded towns than ‘unregulated villages’ is based on misquotation of my study, muddled conflation of guild quotas with labour productivity, and disregard of evidence on guild regulation in villages (Ogilvie 2008, pp. 177-8).

I count 15 refutations.

Technological Innovation

One provocative claim in favor of the guilds suggests that they imperfectly prevented innovation, blocking only labor-saving inventions outright and doing so weakly in any event. Surprising as it may seem, Epstein in one 1998 article even argued that the guilds were themselves creators of new technologies, citing four mechanisms: 1) offering monopoly rents to innovators 2) promoting clustering and spillovers 3) intergenerational idea transfer through apprenticeship 4) spreading ideas geographically by forcing journeymen to travel. The guilds also prevented "harmful" innovations that might have damaged product quality.

Several of these are theoretically intriguing ideas, but they don’t really fit together, let alone work independently. Ogilvie notes that guilds must have been simultaneously weak and strong to a) let labor-saving inventions slip through the cracks and b) enforce monopoly rents to innovators. Monopoly rents aren't inherently pro-innovation and require no/low entry barriers to encourage innovation, which is clearly not the case with the guilds.

The advantages of clustering are obvious in principle: guilds might, like portage sites in the US, serve as coordination devices to choose the location of concentration among multiple viable sites. But the empirical evidence for this point appears to be lacking. Ogilvie compares the technologically moribund German worsted industry with the flourishing and guild-free British and Dutch sectors, whose advance was predicated on the rapid adoption of new foreign techniques on both the product and process sides. And it's no comfort that “only” labor-saving inventions were opposed (even if that were true)—such devices lay at the core of the Anglo-European Industrial Revolution.

Guild regulations, meanwhile, were also unintentionally harmful to innovation. Quality rules prevented changes to the production process, which might have made it harder to monitor. Entry barriers prevented potential innovators from carrying on the trade. Price floors prevented new techniques from undercutting inefficient ones. And long-term apprenticeship in a single trade created vested interests among the masters against dynamism which might render their human capital nugatory. And by preventing non-members from selling, their incentives to innovate were also removed. The list goes on.

In his 2008 article, Epstein abandons the technology-creation thesis, retreating to the informational transmission argument instead. Ogilvie rejoins that apprenticeship was both not universal within guilds and compatible with non-guilded industries, such as English worsted textiles, which were far more innovative than their protected central European counterparts. Outsiders managed to learn how to produce like guild craftsmen and did so successfully enough to attract masters' ire. And the journeymen's mandate to travel was really another entry barrier and was in any case absent in the innovative Netherlands.

One important paper defending apprenticeship as a method of knowledge transfer is De la Croix et al. (2018). The authors construct a model of technological progress in which information sharing occurs person-to-person; thus apprenticeship, which expanded a young worker's circle beyond his family, increased the exposure of individuals to new ideas and techniques. Yet the authors note that apprenticeship is not synonymous with guilds and that the system worked best in Northwest Europe where the state was strong enough to enforce contracts without the anticompetitive influence of informal institutions. Their argument, I believe, can be taken to say that guilded Europe was better positioned than clan-based "everywhere else" to transmit technical ideas, but that the optimal setup was the one prevailing in England and the Netherlands, where apprenticeship coexisted with flexible factor markets.

Many of Europe's premier industrial centers were non-guilded. In 1782, Horn (2012) shows that 85 percent of Normandy's cotton textiles and all of its woolen, stocking, metallurgical, paper, glass, chemical, and ceramics industries were located small no-guild enclaves. Leiden, a leading Dutch textile region, was exceptional both for banning its guilds and for its product and process innovations during the Golden Age. English textile machinery entered Germany via the Rhineland and Saxony, where the guilds lacked power to enforce restrictions.

Political Economy

The major point in the debate on the guilds concerned their role in promoting good governance in urban polities. Robert Putnam (1993) argued that guilds helped to promote civic self-responsibility in Northern Italian city-state (with persistent effects on social norms) while Karl Gunnar Persson deemed them conflict resolution forums between adversarial factions. Epstein, finally, contended that bargaining with guilds helped states to “to overcome multiple coordination failures in the implementation of fiscal, political, military, and economic policy” and reduced “dispersed rent-seeking.”

Epstein's claim is that pre-modern markets were ‘thin’, meaning that coordination of “decentralized agents” would generate positive externalities that would outweigh both the deadweight loss and political costs of rent-seeking. What these externalities are isn't exactly clear. Ogilvie's response is that “particularized” states—which favor specific individuals or factions—failed at delivering the goods (“mobilizing economies of scale, increasing the costs of collusion, and enforcing clearer rules and procedures for contract enforcement”) by comparison with “universalist” ones, where laws applied generally. This seems pretty obvious—why should the characteristics of open-access orders be achievable only by creating a closed-access society?

In sum, the pro-guild case reeks of contrarianism. The arguments in favor flout common sense, empirical evidence, and often mere logical consistency. Worse, Epstein appears to have argued either sloppily or in bad faith, misquoting Ogilvie and misinterpreting her evidence on multiple occasions. He confuses labor productivity with output quotas. His language verges on rhetorical ad hominem attacks on his opponent (e.g. “Ogilvie presents her argument and methodology as novel; in practice they are neither.”) and accusations of “careless reading” and “myopia.” Some of the arguments are interesting and provocative, but Epstein rarely presents sufficient empirical evidence to show that his mechanisms actually, well, worked. I feel pretty sympathetic to Ogilvie’s conclusion:

Epstein’s article… is a tissue of unsubstantiated assertions underpinned by startlingly inaccurate criticisms of my German case study, quotations from cognitive psychology textbooks, and references to a small circle of like-minded believers (Ogilvie 2008, p. 180).

The most compelling part of Epstein’s 2008 critique is to stress that the single German case study may not generalize to other regions and that until a more comprehensive picture emerged one could (if unjustifiably) treat the Anglo-Dutch cases—which he favored—as representative. In response, however, Ogilvie later wrote The European Guilds (2019), a sweeping compilation of more than 17,000 observations of guild behavior in 22 modern countries over 8 centuries. The book's comparative perspective presents a powerful confirmation of the generalizations from the Württemberg case study. Markets were near-ubiquitous in early modern Europe, and the constant efforts by guilds to restrain them for narrow advantage spoke against them—if markets didn’t work, why did guilds pay so much to stop them? And if guilds imposed little practical restraint on economic activity, why did artisans forfeit fees (150-300 days’ wages) and time to join?

Do Guilds (or the lack thereof) Explain Anglo-Dutch Growth?

The case against guilds is compelling. Guilds created market frictions, blocked technological progress, were indifferent toward educating workers, and mired societies in “particularized” regimes in which favoritism and protection assigned to interest groups limited competition and development. They existed not because they were efficient, or because their net benefits outweighed their costs, but because they were vehicles by which small bands of economic elites could effectively lobby rulers for privileges. So how much did this drag on growth? And how much of the Little Divergence do they explain?

Anglo-Dutch guilds began to weaken during the sixteenth century. In England, guilds vanished from some of the older towns (which had never granted local monopolies), had little sway over the countryside (to which proto-industry escaped), and never formed in the new industrial towns of Birmingham, Manchester, Leeds, Halifax, Sheffield, and Wolverhampton. Established centers “gradually lost their ability to conduct quality inspections, regulate production techniques, compel apprenticeship, prevent women’s work, or enforce entry restrictions.” In 1689, only a quarter of England's 200 towns had any guilds at all, and as we've discussed above some crucial industries in guilded towns weren't organized themselves.

Patrick Wallis (2008) shows that London's guilds failed to enforce the completion of apprenticeship contracts by the late seventeenth century, but that apprenticeship itself thrived nonetheless. This was due to the relatively even distribution of training and payments throughout the life of the bond, which ensured that individuals could drop out when they felt they'd received sufficient training without costing much to their masters. Thus in England, the human capital benefits of guilds were both be separated from the institutions themselves and more widely diffused in their absence.

The Dutch Republic presented a similar, if more complex, picture. The number of guilds increased during the seventeenth-century Golden Age, but primarily in local industries (rather than export trades). Ogilvie quotes Soly to the effect that “[t]he overwhelming majority of the textile producers in the United Provinces, employers and employees alike, operated outside the corporative context.” Leiden textile workers were organized in neringen to which all producers inherently belonged, informal institutions supervised by municipal authorities. Similar frameworks were present in the Republic's other famous industrial towns—Amsterdam, Haarlem, Delft, and Gouda.

France and Italy were intermediate cases; in France, Lyon, an avowedly “free” city, housed a dominant silk industry, and other towns like Roubaix lay outside the guilds' scope. In Italy, Vicenza played a similar role. Germany and Scandinavia, however, had “pervasive” guilds.

Why were guilds stronger in some places than others? One answer offered by Ogilvie is the density of and competition in a country's urban network—the less urbanized a country, the more that each town represented a local monopoly over which protection was possible. Many nearby towns meant overlapping hinterlands, which in turn prevented any particular guild from controlling the countryside. In England and the Netherlands, the absence of city-states allowed competing regional interests to neuter each other in the national assemblies. Dutch governments always exercised greater control over the guilds, using them as tools for advancing the general public interest rather than submitting to their lobbying behavior.

From the sixteenth century on, the weak-guild Low Countries and England began to diverge in per capita GDP from intermediate and strong-guild regions. While it's unwise to make too much of these comparisons, it's important to note that no country with practically strong guilds—that is, where guilds actually exercised control over their trades and political influence—experienced rapid economic growth prior to the nineteenth century. Guild abolition immediately preceded industrialization in several European countries, including France and northern Italy, and Napoleon’s conquests extended the movement into the Rhineland, the Netherlands, and southern Italy.

The abundant urban evidence in The Economics of Guilds supports the aggregate view. But without resorting to econometrics, it's hard to get a hold of what “the effect” of guild strength truly was. Guilds are obviously highly endogenous. Cities and trades experiencing economic decline, like those of Holland after 1680, would invite increasing guild power as a form of protectionism—suppressing rural industry and foreign competition to defend profits. Indeed, it feels like there may be some low-hanging fruit for a study of European city growth based on this or similar data, a la “Resetting the Urban Network” or “Coal and the Industrial Revolution”—i.e. a difference-in-differences design comparing strong- and weak-guild regions or an IV.

Kelly et al.’s claim—that in weak-guild England, the availability of apprenticeship outside guild jurisdiction provided an abundant supply of skilled workers to nascent industries and transmitted knowledge to a wider public—does ring true, in this light. But we're no closer to an explanation for England's separation from the Dutch Republic, whose rise appears to have centered around notably free cities/regions like Leiden and Hondschoote. In comparison with central Europe, however, the story does appear to be plausible.