The price of human capital in a pre-industrial economy: Premiums and apprenticeship contracts in 18th century England

Available online at www.sciencedirect.com

Explorations in Economic History 50 (2013) 335 – 350 www.elsevier.com/locate/eeh

The price of human capital in a pre-industrial economy: Premiums and apprenticeship contracts in 18th century England☆ Chris Minns⁎, Patrick Wallis1 Economic History Department, London School of Economics and Political Science, Houghton Street, London WC2A 2AE, United Kingdom Received 20 April 2012 Available online 13 March 2013

Abstract Training through apprenticeship provided the main mechanism for occupational human capital formation in pre-industrial England. This paper demonstrates how training premiums (fees) complemented the formal legal framework surrounding apprenticeship to secure training contracts. Premiums varied in response to scarcity rents, the expected productivity of masters and apprentices, and served as compensation for the anticipated risk of default. In most trades premiums were small enough to allow access to apprenticeship training for youths from modest families. © 2013 Elsevier Inc. All rights reserved. JEL classification: K31; J24; N33; O15 Keywords: Apprenticeship; Training; 18th century England

1. Introduction The price of education and training, and how these costs are distributed within a society, are major determinants of the extent of human capital formation. The ☆ We thank STICERD/LSE Annual Fund for financial support, Steve Broadberry, Frank Lewis, Leonard Schwarz, conference participants at Gettysburg, Tuebingen, and seminar participants at Essex, Paris School of Economics, and Zurich for their comments, and Cliff Webb, Ian Galbraith, Mark Merry and the Records of London’s Livery Companies Online project for kindly supplying data. Part of this research was completed while Minns was a Weatherall Distinguished Fellow at Queen's University, Canada. ⁎ Corresponding author. Fax: + 44 207 955 7730. E-mail addresses: [email protected] (C. Minns), [email protected] (P. Wallis). 1 Fax: +44 207 955 7730. 0014-4983/$ - see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.eeh.2013.02.001

advance of mass schooling was most rapid in nations where the cost of educating children was lower relative to incomes, and the financing of public schools reduced the burden of costs faced by poorer members of society. The evidence suggests that this mattered for growth: the early leaders in human capital formation through mass schooling were among the most precocious national or regional economies prior to 1900 (Lindert, 2004; Go and Lindert, 2010). The importance of human capital formation did not begin in 1800, however. A number of economic historians have suggested that high levels of occupational skill were one of the root causes of industrialisation in England and Europe (Mokyr, 2009; van Zanden, 2009; Humphries, 2003). Apprenticeship inevitably plays a major role in these discussions: it was the leading formal source of vocational skills outside the agricultural sector in pre-

336

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

modern England, with roughly nine percent of English male teenagers having entered apprenticeships in London alone around 1700 (Minns and Wallis, 2012). Financing an apprenticeship prior to 1800 posed many of the same problems faced by families wishing to educate their children through formal schooling after industrialisation. One of apprenticeship's potential advantages is that it can allow youths to finance training by working for their employers at wages below their marginal product. Apprenticeship therefore reduces the regressive effects that credit constraints impose on people's capacity to invest in their human capital. Industrial apprenticeships in the nineteenth and twentieth centuries often operated on these terms (More, 1980; Smits and Stromback, 2000). In the seventeenth and eighteenth centuries, indentured servitude utilised a similar system: English youths obtained passages across the Atlantic by committing themselves to work for an employer for a period thereafter (Galenson, 1981). It is striking, therefore, that English apprentices in the eighteenth century were often required to make an up-front payment, known as the premium, to their master in order to obtain a position. The existence of premiums is well known to historians, who have often suggested that they were large and ubiquitous. As a result, premiums have been interpreted as lowering the extent of economic mobility through apprenticeship. If this was the case, it offers a potentially powerful counter-argument against suggestions that the relative openness of apprenticeship, and by extension of guilds and skilled occupations, was one of the distinguishing features of the European economy (Epstein, 1998). Large, uniform premiums would reinforce more pessimistic accounts of apprenticeship as a socially exclusive device that preserved rents for insiders (Ogilvie, 2004). Premium size is thus an important variable for wider economic performance — if it presented a significant barrier to training, this would have implications for productivity, economic mobility and the extent to which aptitude and opportunity were matched (Ben-Amos, 1994; Brooks, 1994; de Munck, 2007. See also: Ogilvie, 1997, 2004). Premiums also have significant implications for a second element in ongoing debates about apprenticeship: whether masters and apprentices were able to write viable, self-enforcing contracts. Several studies have argued that apprenticeship contracts in Paris, Montreal and Antwerp were tightly specified in order to reduce opportunism (Hamilton, 1996; Kaplan, 1993; de Munck, 2007). No parallel study exists for English apprenticeship, however, and recent work has focused on external enforcement mechanisms (Humphries,

2003; Minns and Wallis, 2012; Wallis, 2012). Indeed, despite the significant role attributed to apprenticeship in pre-industrial economic development, there is relatively little systematic evidence about the incidence and level of premiums that English apprentices actually paid, and no detailed analysis of what factors determined the level of premiums in this period. In this paper we explore the price of apprenticeship training in England in the eighteenth century, the effect of fees on access to training, and the role of fees in permitting contracts to be agreed in an environment with limited institutional flexibility. Using a large sample of premium records from the eighteenth century, we show that the financial barrier presented by premiums has been overstated, varied widely between occupations and individuals, and was in many cases dwarfed by the cost of starting a business following a successful apprenticeship. We find that premiums fit well with a standard human capital model, expanded to incorporate the risk of early departure. They reflected variations in the anticipated costs and benefits of individual training contracts, including apprentices' expected productivity, and the likely availability of economic rents following training. However, we also suggest that one of the main purposes of premiums appears to have been to solve the problem of potential holdups in the training market due to the high risk of default. Premiums allowed masters to accept apprentices readily and invest in training their apprentices despite high rates of attrition. 2. Training contracts in pre-industrial England: theory and history As a system of on-the-job training, apprenticeship involved skill transmission over an extended period. For apprentices and masters to be willing to enter into contracts, both parties must expect to receive positive net benefits as an outcome of training. At first glance, apprenticeship in pre-industrial England fits well with a stylised model of general human capital acquisition (Becker, 1964): apprentices had greater productivity and earnings capacity after training, and masters had ample opportunity to recover costs from trained apprentices over long terms of seven years or more. However, the long-term contracting required in this model presents a fundamental problem for all parties. The specification of apprenticeship contracts is problematic, with both parties struggling to commit to incomplete contract over a long period (Grubb, 1997; Humphries, 2003). Masters may shirk on training or wages. Apprentices may quit prematurely once the

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

wage they could obtain in the outside labour market exceeds that received from the master. The price of training in pre-industrial England can be analysed through a simple model of apprenticeship. In a competitive equilibrium, the price paid by apprentices to train will equal to the costs borne by masters who provide training. Eq. (1) describes the master's problem in this market. Apprentices who secure training receive a wage (subsistence payment) (w) in every training period, and may receive a post-training benefit on completion (B). Masters receive the product of apprentice labour in each period (v), pay training costs (c) in the form of instruction or supervision, and may receive a training premium (P) from the apprentice. We also include the probability of persistence in the indenture (ρ). With r as the discount rate and T as the length of the indenture, Eq. (1) illustrates equilibrium in present value terms2: T X t¼0

ρt

T X vt −ct T þ ρ B ¼ ρt þ P: t ð1 þ r Þ ð1 þ r Þt t¼0

wt

ð1Þ

Eq. (1) outlines the key parameters of the apprenticeship indenture that could, in theory, be varied to generate successful matches between apprentices and masters. As is sometimes observed in present-day apprenticeship, training wages (w) can be varied to offset planned training inputs (c) and apprentices' anticipated productivity (v). The length of the indenture term (T) can also be adjusted; several historical studies of indenture contracts find that productivity and length of term were inversely correlated. 3 Post training incentives (B) can be used to make contracts self-enforcing. Today, this might include ongoing employment in the firm providing training. 4Historically, part of the apprentice or servant's compensation was sometimes received on completion (Hamilton, 1996; Galenson, 1981). While Eq. (1) suggests multiple avenues through which training contracts could be varied to clear the market, the economic environment of pre-industrial England presented additional challenges. First, the institutional setting of English apprenticeship gave masters and apprentices much less flexibility over indenture arrangements than was available in other 2

This model is an extended version of that found in Hamilton (1996). 3 See Kaplan (1993) and De Munck (2007) for apprenticeships in Paris and Antwerp, Hamilton (1996) for Montreal apprentices, and Galenson (1981) for indentured servants bound for the Americas. 4 Market imperfections may give training firms some monopsony power in setting wages for generally trained employees: see Acemoglu and Pischke (1999).

337

historical and contemporary contexts. Although indentures were private contracts between individuals – apprentices (and their families) and masters who were in effect proprietors of small enterprises – there was little freedom to vary most elements.5Local and national laws defined the main features of apprenticeship in England until 1813. Unlike most other parts of Europe, this legal framework gave little scope for contract variation to match the distribution of expected costs and benefits of training. Nationwide, the minimum contract duration was seven years. The payment of money wages to apprentices was illegal in many places (including London), though some did receive payments or perquisites. A second challenge to achieving optimal levels of contracting for apprenticeship is the overwhelming evidence of high levels of attrition among apprentices. Recent studies of late 17th century London and Bristol suggest that the majority of apprentices did not complete their term, a finding that is reinforced by earlier studies of completion and citizenship (Minns and Wallis, 2011; Ben-Amos, 1991; Wallis, 2008). High levels of attrition present a serious challenge to standard apprenticeship arrangements. If the outlay of masters (w + c) was front-loaded relative to the recovery of surplus from apprentices (v), masters risked being unable to recover their investment in apprentices if the complete indenture term was not served. Masters could alternatively contract on the anticipated term for a given rate of attrition (1-p). One way to compensate for early departure would be to lower the outlay on apprentices by reducing w or c. It is unclear how realistic it was for masters to do this. Lowering w below subsistence would affect productivity. Reducing c raised the risk of apprentice default.6Alternatively, masters could only accept apprentices with a high initial v. The use of trial periods suggests that masters sought information about an apprentice's productivity.7However, screening was costly in an early modern environment, and there is little evidence that training was constrained by masters being especially choosy about the quality of their apprentices. Another alternative would be to alter the sequence of training investments and wage payments (Wallis, 2008). By placing a significant share of the 5

Some apprentices and masters did make supplementary agreements over the costs of clothing and maintaining apprentices, bonds for the apprentice's honesty, and provisions for end of contract payments. 6 There is evidence that masters economised on training costs by providing little direct instruction (Wallis, 2008). 7 See Minns and Wallis (2011) for evidence on the presence of trials.

338

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

outlays towards the end of the contract, masters would limit losses from apprentices who quit early, while providing an incentive for others to remain attached to the indenture. Altering the sequence of training investments would have implications for apprentice output over the term, and the skills acquired by apprentices who completed indentures as well as those that left early. Evidence that apprentice attrition was common poses an obvious challenge to most explanations of apprenticeship's success, which focus mainly on contract enforcement by formal institutions (van Zanden, 2009; Epstein, 1998). Pre-industrial England did offer distinctive post-training incentives to apprentices that some have suggested made contracts self-enforcing (Humphries, 2003). Most importantly, completing an apprenticeship was a legal requirement before practicing a trade (Davies, 1956). Apprenticeship also conferred local economic privileges through guild membership and citizenship in incorporated cities, and positive effects on reputation. 8 Masters and apprentices who breached contracts could be pursued at law. However, the impact of these incentives is questionable, particularly for apprentices who did not envisage establishing an independent business. They left open several important outside options for ex-apprentices, including agricultural and military employment, as well as work as journeymen.9In practice, the numbers of apprentices who were allowed freely to exit their contracts by local magistrates and courts far exceeded the number coerced into continuing service. Evidence from London in the seventeenth and early eighteenth centuries shows that the city's Lord Mayor's Court cancelled thousands of apprenticeship contracts virtually on demand. Around one in ten London apprentices abandoned their contracts in this way, even when their masters objected. Fragmentary evidence from other courts suggests that similar practices occurred elsewhere in England; the penal application of the law of master and servant against employees appears to have been a phenomenon of the nineteenth century (Wallis, 2012). Limited contractual flexibility and high rates of apprentice attrition suggest that training premiums (P), may have played a particularly significant role in clearing the market for training, compensating both apprentices and masters for perceived differences in productivity and the likelihood of contractual default. 8 Humphries (2009) observes an effect of completion on income among early industrial apprentices. Rights to poor relief are sometimes cited, but in law serving 40 days of an apprenticeship conferred rights. 9 Journeywork technically required a completed apprenticeship, but this was variably enforced in different trades.

There were no restrictions under the statues of English apprenticeship law regarding whether training premiums were charged, and what their size might be. As we will show, both the size of premiums and the frequency with which they were charged varied widely between and within guilds and occupations, and even among apprentices indentured to the same master. Returning to the model of apprenticeship described by Eq. (1), a simple re-arrangement can be used to derive an expression for the determinants of training premiums: P¼

T X t¼0

ρt

wt −ðvt −ct Þ ð1 þ r Þt

þ ρT B:

ð2Þ

Eq. (2) suggests three broad explanations for the size and frequency of training premiums. The first is that training premiums were used to compensate masters for the early, unscheduled departure of apprentices before their term of indenture was complete. While there is some evidence that masters pursued alternative solutions to the problem of apprentice attrition, the use of up-front training premiums as side payments would have allowed masters to charge apprentices in advance for the direct costs of training, and in doing so provide advance compensation for potential attrition. Rather than limiting training to the most highly committed under conditions of minimal expenditure on training costs, the use of premiums offers a potential solution to the classic problem of credible commitment in firmprovided general training without creating incentives for masters to further constrain training opportunities beyond those imposed by guilds.10 If this explanation for training premium size is valid, premiums should reflect the anticipated probability of attrition (1 − ρ) in any particular contract. Some of the reasons for attrition, such as mortality, and masterapprentice match quality, were difficult to observe in advance. But attrition also depended on the intentions of 10

An alternative solution to apprentice attrition would be to require apprentices to post a bond when commencing training. Bonds for the apprentice's ‘truth’ or honesty were used where embezzlement was a risk. However, there were three practical difficulties in applying this arrangement to apprenticeship contracts: it would require that apprentices have credible local underwriters, which many migrant apprentices would be unable to identify; with attrition rates of 60% it would be difficult for any apprentice to secure guarantors; and finally, proving that the apprentice's faults, not the masters, led to departure was inherently difficult (see Wallis, 2012). Moreover, masters may have been indifferent between scenarios featuring bonds and low attrition versus premiums and high attrition, but apprentices who could afford to pay and did not seek to be locked in for seven years would prefer the second alternative.

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

masters and apprentices, which may have been partly predictable. Unfortunately, the sources recording training premiums do not provide any direct indications about whether an apprentice left their indenture before completion of term.11Other evidence for the late 17th century, however, confirms that attrition occurred frequently. In a samples of London and 1300 Bristol apprenticeship indentures linked to household tax records, over half of contracted apprentices were no longer living with their master during the prescribed term of indenture (Minns and Wallis, 2012). Early departure was commonplace in both elite occupations and manual trades. Evidence from the city of Bristol shows that over a third of grocers' apprentices and over half of merchant apprentices were absent from the master's household, compared to a rate of about forty percent for coopers apprentices. Eqs. (3) and (4) describe the apprentice's problem in entering or leaving an apprenticeship. These decisions depend on the expected present value of future income during training and following completion (Yc) relative to the expected present value of income in alternative employment (Yo). Yc and Yo are functions of match quality (Mc and Mo), apprentice skill (v) and remuneration of skill in the local labour market, and the resources (R) that the apprentice brings to post-training activity: Yc ¼ f ðMc; v; RÞ

ð3Þ

Yo ¼ f ðMo; v; RÞ:

ð4Þ

  Yc dv and ∂ Yo dv are the return to “skill” in both markets. Apprentices' resources (R) include family specific attributes, such as financial and social capital, and potential inheritances. Social capital and family resources should affect both the inside and outside options, but high transportation costs and poor communications imply  that these  would matter more for Yo than for Yc, with ∂ Yo dR N ∂ Yc dR .12 The inclusion of R in Eqs. (3) and (4) implies a significant deviation from simple, competitive labour markets. This appears plausible given post-training opportunities in an economy without large firms. Establishing a business required significant capital unavailable to the credit-constrained majority; indeed, many apparently “failed” apprentices who quit their ∂

11

Some indirect evidence on departure exists in a small minority of guild records that indicate apprentices who became freemen. 12 See Minns and Wallis (2012) for supporting evidence on this point.

339

contracts may have moved away to positions that made direct use of family resources. Accounting for the costs of exiting an apprenticeship (Cx), youths should leave indentures if Yo − Yc N Cx for the balance of the term. Initially, apprentices would have better information about Yo than Yc. As they learned more about Yc over their term, departure would become attractive for some apprentices with better outside options. Masters would know less about their apprentices' outside option, but received some signal about R from the occupation or social origins of the apprentices' father and their location. All else being equal, masters would anticipate a higher probability of departure for apprentices with better relative outside options (Yo − Yc) at the outset, and charge higher premiums as a consequence.13Table 1 provides some indirect evidence that this expectation was reasonable. Migrant apprentices with origins in the provincial elite (sons of gentlemen and yeoman) or from provincial mercantile families were significantly more likely to abandon their indentures than local London youths of similar background. These youths are among those most likely to have significant social and financial capital away from the site of training.14 Risk averse masters should also condition premiums on uncertainty surrounding apprentices' productivity. This would imply masters charging higher premiums when it was more costly to collect information about v, R, and Mc. If apprentices were paying larger premiums in order to compensate masters for the risk of their departure, what was to prevent masters from defaulting on their own commitments to train the apprentice and appropriating the premium for themselves? Here a different institutional mechanism provided an important complement. Premiums were partially returned when contracts ended prematurely with the division depending in large part on the proportion of the term that had been served. Magistrates and city courts enforced this division if it was not privately agreed (Wallis, 2012).15 13

Apprentices from poor families suggest ambiguous predictions. At face value, they have fewer outside options, but conversely, the lack of family capital may have reduced the benefit of completed apprenticeship by reducing the likelihood of establishing an independent business. 14 Differences in departure costs (Cx) should also matter for this assessment. We do not have firm evidence on this, but expect that the costs of running away may have been lower for migrant apprentices, who could find family members to take them outside of their training location. 15 Apprentices sometimes received a modest exit payment, but these were small relative to premiums and became rarer over time (Grubb, 1997). Unfortunately our source material provides no evidence on completion payments.

340

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

Table 1 London apprentice attrition by occupation group, 1688–96. % Observed with master Father occupation group

London origin

Migrant origin

Difference N

Gentleman Professional Distribution and sales Yeoman Craft Service Labourer

53 30 60 64 46 20 25

45 49 41 47 55 55 86

8 − 19 19 b 17 −9 − 35 a − 61 a

83 63 92 121 248 47 11

Notes: Derived from data used in Minns and Wallis (2012; Table 4). a Statistically significant at the 5% level. b Statistically significant at the 10% level.

A second explanation for premium size is that this should respond to signals about the productivity of apprentices, their expectations about treatment while contracted, the quality of training likely to be provided by masters, and the cost to the master of providing training.16This interpretation is supported by most theories of premium payment and premium size put forward by historians, which generally link premiums to the net costs or returns from training (De Munck, 2007).17 Our final explanation is that premiums reflect the scarcity value of training. There is some evidence that guilds sought to create scarcity, by restricting the number of apprentices their members could train, often to one or two at a time. This “soft cap” on availability would sustain high premiums when demand for training was high. Limiting flows into apprenticeship may have imposed barriers to the flows of ideas and production techniques outside of the guild, thus ensuring economic rents to members. Despite such measures, however, apprentices and their families generally faced a range of opportunities, and large numbers of masters did engage in training (Leunig et al., 2011). In London alone, 74 different master

16 Highly productive masters with significant opportunity costs associated with time spent training apprentices may have sought a (larger) fee as compensation. 17 Prestige (Kaplan, 1993, p. 449; Brooks, 1994; De Munck, 2007, p.42) and economic rents from guild or city privileges (Earle, 1989) are often included among the returns to training. Reith (2007, p. 183) suggests that fees were only charged for weaker youths by Augsburg bakers. De Munck (2007) claims that fees reflect teaching costs. In addition, it is likely that the contribution of otherwise similar apprentices to the master's output varied between occupations and activities. Apprentices were likely employed as factor by merchant masters, but saw less direct use in skilled services early in the indenture.

apothecaries are recorded as accepting premiums to train between 1725 and 1750, while 86 different masters took on haberdasher apprentices over the same time frame. Guild members were not the only suppliers of training. No restrictions applied outside guild towns, and by the eighteenth century even in guild towns apprenticeships were increasingly offered by people outside guilds. Nonetheless, masters in trades where guild constraints were effective could potentially use premiums to capture economic rents, while their apprentices might expect to gain access to some of these rents after training. In terms of Eq. (2), scarcity rents would be associated with positive values of B that apprentices would be willing to pay for when indentured.

3. New evidence on apprenticeship: premiums and indentures Our evidence derives from two sources recording the indenture characteristics of large numbers of apprentices, masters and the premiums paid. The first details 300,000 apprenticeships on which masters paid Stamp Tax. From 1710 to 1804, apprenticeship premiums were taxed at 2.5% on fees up to £50, and 5% over £50.18The Commissioners of Stamps recorded the payments, noting the name, address, and trade of masters, the name and sometimes the family background of the apprentice, and the date of indenture.19We have digitised a typescript abstract of the registers from 1710 to 1774. The Stamp Tax also allows us to reconstruct the training and premium histories of around 9000 masters by linking masters' names, places and occupations. We then compute their total number of apprentices and average premium, and have identified each apprentice's position in their master's training history. Given that information on apprentices' father's occupation and place was recorded intermittently, and that some apprentices cannot be matched unambiguously to a master, the base sample we use for regression analysis contains just under 30,000 observations. We also linked a sub-sample of just over 4000 apprentices to baptismal records in the International Genealogical Index (IGI), for which we calculate their age when apprenticed.20

18

After 1804 this tax structure was replaced by a simpler banded tax. The records of the Commissioners of Stamps continue to record late payment of premiums until 1811. 20 The linkage is discussed in Wallis et al. (2010). 19

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

The Stamp Tax data do have flaws. The records are affected by lags in registration and occasional losses where provincial records were not properly registered: for a few years almost no premiums are recorded. The Tax records only record cash payments, and many families also covered part of the apprentices' living expenses, most commonly by supplying their clothing. Perhaps the most important shortcoming is that they omit apprentices who paid no premium. The second series consists of apprenticeship contracts registered by eight London livery companies. After the introduction of the Stamp Tax, some guilds began to record premiums alongside the information they gathered about apprenticeships. We use the continuous series of contracts with premiums recorded by the Clothworkers' and Stationers' Companies, and samples from the early, mid and late eighteenth century for the Apothecaries', Blacksmiths', Grocers', Plasterers', Turners' and Vinters' Companies (attached to existing complete series of contracts, Webb 1994-). A crucial aspect for our analysis is that guilds are often noted when no premium was paid. Guild data, however, is often unclear about masters' occupations. Although many no longer followed the guild's nominal trade, precise occupations are rarely recorded. Our data covers only a minority of guilds, and many apprentices were trained (and paid premiums) outside of the guild system in this period. Matching apprentices found in both the Stamp Tax and company records provides further insight into the

341

strengths and weaknesses of these two sources. We are able to find a strong and unique match for about 65% of Stamp Tax apprentices in our Company sample. Among these, we find no evidence that masters reported lower premiums for tax purposes — when a London apprentice can be found in both sources, the premium paid is identical in over 84% of cases. While all these records have a premium recorded in the Stamp Tax, this was not always the case in guild records. We find that a small number of cases (about 5% of the matched sample) do not provide any information about premiums in the guild record, while a smaller share (about 1%) include explicit indicators that no premium was paid. This is perhaps not surprising given that guilds only record premiums patchily, and had no reason to check the accuracy of the sums reported to them. In sum, the comparison exercise suggests that the guild records modestly overstate the share of apprentices who did not pay a premium. The share of unrecorded premiums in the guild records reduce the sample size under analysis, but missing premiums do not appear to be correlated with premium size (when recovered from the Stamp Tax), or the characteristics of masters and apprentices. 4. Premiums and access to apprenticeship The Stamp Tax data suggest stable flows into apprenticeship between the 1710s and the 1770s (Fig. 1). The annual fluctuations apparent in the second

Fig. 1. Stamp Tax indentures, 1710–1773. Notes: Stamp Tax sample; all observations where master county was recorded. See text for further details.

342

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

Fig. 2. Apprenticeship premiums, 1710–1773. Notes: Stamp Tax sample; all observations where master county was recorded. See text for further details.

quarter of the century mainly reflect record survival. We find no sign of a decline in apprenticeship during the eighteenth century, despite falls in the numbers of apprentices registered by guilds (Minns and Wallis, 2009; Snell, 1996; Walker, 1986).21Premiums rose gradually in London until the 1740s; outside of London payments were much lower (Fig. 2).22Much of the difference between London and the provinces was due to the concentration of prosperous trades that charged high premiums within the metropolis.23 Historians have suggested that premiums were large and ubiquitous. For apprentice merchants, Grassby suggests that large sums of about £100 in 1700 and £200 to £300 in the early 19th century were usual (Grassby, 1995, pp. 67–9). In trades, Brooks suggests that premiums of £20 to £30 were common (Brooks, 1994). Such figures are often based on fragmentary data, usually from selective London sources (for example, Earle (1989), p.84).24We find that premiums

21

The rate of entry into apprenticeship also appears fairly constant over the period (Minns and Wallis, 2009). 22 Fig. 2 presents nominal apprenticeship premiums. Adjusting for the trend in consumer prices (Allen, 2001) implies a gentle rise in real terms. 23 Average premiums in the city of London were about 35% greater than in the largest 11 provincial towns in England (de Vries, 1984). 24 Van der Beek (2010) offers more evidence of premium size from a sample of Stamp Tax premium payments.

were generally more modest. In most crafts and trades, including clothing, footwear, textiles, and metal manufacturing, fees below ten pounds were typical (Table 2). Professionals typically charged around £50, and there was more heterogeneity in prices for these occupations. Despite large difference in mean premiums

Table 2 Apprenticeship premiums, selected occupations, 1710–1773.

Food industries Clothing Footwear Textiles Wood industries Iron and steel manufacture Building and construction Other services Professions

Mean

Median

Standard deviation

N

10.0 11.5 6.5 15.7 15.6 9.4 10.9 16.1 73.6

9 7 5 5 10.5 5 10 10 52.5

13.5 18.1 8.4 35.7 17.5 19.4 12.5 33.0 60.6

21627 32560 27436 19863 13504 11649 34684 8655 16767

Source: Premium payments recorded in the Society of Genealogists typescript index to the Register of Duties Paid for Apprentices' Indentures 1710–1811 (The National Archives, IR1). We use the full population of all surviving indexed indentures between 1710 and 1774. In addition to premiums, the registers record the names of apprentices and masters, the location of the indenture, master occupation, parent name (usually the father), and often the occupation of father. Notes: Occupation groups are based on Wrigley's P.S.T. codes to the second digit (Wrigley, 2004).

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

343

Fig. 3. Premium distributions, selected occupations. Notes: Distributions are kernel density estimates of positive premiums of 100 pounds or less. The Epanechnikov kernel is used, with a bandwidth of 4.

between occupations, the distribution for specific occupations overlapped substantially (Fig. 3). London guild records tell the same story as the Stamp Tax data (Table 3). While some aspirant merchants or professionals paid large sums, they made up a small minority of apprentices. In many of London's craft occupations, a premium of 10 pounds was typical. Nationwide, only 8% of apprenticeships cost over 50 pounds, and 5% exceeded 100 pounds; even in London, the figures were only 19% and 6%. London guilds also recorded when no premium was paid. In relatively low prestige guilds (Blacksmiths, Plasterers, Vintners) only a minority paid premiums. The few companies which noted the master's occupation mainly contained prosperous trades, but even here we see large differences: the majority of apprentices in several craft occupations, such as plasterers, blacksmiths, and butchers paid nothing. The probability of paying a premium was higher in occupations where premiums were larger, but even here a third of apprentices training as druggists or grocers did not pay. Even if we adjusted for under-registration of premiums in guilds, the wide range among companies and occupations would remain. Were premiums a barrier to entry? While premiums were smaller than the literature suggests, they were not trivial sums. Unskilled workers in the building sector in provincial England earned roughly £12 a year. These families would require several months' income to pay

£5 to £10 for an apprenticeship with a mason or weaver.25 The larger premiums paid to access some occupations equalled several years' income. For youths without family backing, raising a premium would have been an onerous task. It would take roughly two years of agricultural service to raise the funds to pay a premium of £5.26Compared to alternative human capital investments, premiums were expensive. Schooling a child, for example, cost around one pound per annum.27There is little evidence that families used credit to pay premiums. In the 18th century, the English credit market supplied short term credits or secured loans, not the unsecured long-term loans required for training (Muldrew, 1998). Premiums did create a threshold that poor families struggled to cross. The effect is most acute near the bottom of the income distribution. Labourers made up around 24% of workers in this period (Shaw-Taylor, 2010), but supplied just 4% of those apprentices who paid premiums. Mercantile masters recruited primarily from the top of society, and the majority of apprentices 25 Provincial wages are the average for building labourers, 1710–1772 (Clark, 2005). Days worked are from Voth (2001), p. 1076 (1750 estimate). 26 Kussmaul (1981): 38. Average wage for servant in husbandry under 15 was £2 10s, rising to £5 15s at 20. 27 At a fee of 6d per week, and 45 weeks teaching a year, school would cost £1 2s 6d per annum. Humphries finds an average of three to four years of schooling among eighteenth century autobiographies (Humphries, 2009, p. 314).

344

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

Table 3 London training premiums, companies, and selected occupations, 1710–1800. % Average Sd Paying premium Companies Apothecaries Blacksmiths Grocers Plasterers Stationers Turners Vintners Clothworkers All companies Occupations Bookbinder Bookseller Druggist Grocer Haberdasher Instrument maker Plasterer Printer Stationer Blacksmith Turner Dyer Tailor Butcher Calenderer Watchmaker Engraver

84 23 53 28 46 52 30 36 33

73 11 88 12 36 13 21 36 36

43 47 69 64 79 59

15 90 203 101 61 16

28 35 51 17 64 37 56 29 14 57 74

8 21 79 7 12 12 14 12 8 17 33

36 17 91 15 44 15 36 57 54

Median P10 P90 N

Clothing Footwear Textiles Other Professions Services Median premium

63 6 60 7 20 10 11 20 15

40 3 5 4 5 4 5 5 5

14 10 82 70 112 200 94 92 39 50 19 10

4 11 100 25 21 5

4 20 80 3 7 10 12 8 8 16 28

Table 4 Parental background of apprentices in selected occupations, 1710–1772.

6 19 50 7 10 10 10 11 5 15 25

105 585 21 2832 200 705 20 683 100 4428 26 816 30 1628 84 5862 100 15403

30 200 315 200 100 30

879 714 59 200 119 299

5 15 4 40 10 190 3 10 5 25 5 20 5 30 5 21 3 14 5 30 10 60

204 2691 905 53 87 251 100 169 545 194 307

Source: Sample of eight London companies. For the Stationers and the Clothworkers we have premium data from 1711 to 1800: McKenzie, (1978); Centre for Metropolitan History, (2012). For the Blacksmiths, Grocers, Plasterers, Turners and Vintners we use three samples from 1711 to 30, 1745 to 55 and 1780 to 90. For the Apothecaries premiums are only available for the latter two periods. The source records premiums paid or their absence, the names of apprentices and masters, parent name (usually the father), and often the occupation of father. Premium data: Guildhall Library, London, MS MS8207/ 1(Apothecaries), 2886/4-5 (Blacksmiths), 11,598/2-3 (Grocers), 6122/ 3-4 (Plasterers), 3302/2-3 (Turners), and 15,220/2-3 (Vintners). Indentures: Webb 1994-. Notes: N is the total number of observations, not the number of observations for which we have reported premiums. The descriptive statistics are calculated for observations with positive premiums.

5

% father occupation group Primary 25.2 31.5 Manufacturing 40.1 42.1 Distribution 3.2 1.5 Sales 3.5 2.3 Labourer 10.7 9.9 No occupation 0.1 0.1 Service 5.4 10.0 Professional 3.6 1.2 Gentleman 8.2 1.6 N 3295 2235

5

10

52.5

22.9 42.6 4.6 5.4 4.6 0.3 4.7 4.2 10.6 2569

17.0 38.0 4.7 5.7 1.7 0.1 14.3 6.0 12.5 1845

7.6 15.7 7.8 4.6 0.3 0.1 5.4 18 41.4 2438

Notes: Stamp Tax sample; see notes to Table 2 for further details.

were often only a small part of the investment required to develop a career as an independent artisan. The opportunity costs associated with foregone income dominated premiums for most would-be craft apprentices.28Opportunity costs do not require up-front financing as premiums did, but poorer families may have struggled to subsist without the earnings of teenage boys. A further cost that also often dwarfed premiums was that of the capital required to set up an independent business after training was completed. In prosperous trades, this was perhaps ten or twenty times that of the premium levied (Campbell, 1747, pp. 331–340). The evidence suggests that premiums did limit access to training in England, however the barrier they created was smaller, and more marginal, than was previously thought. 5. Premiums and indenture characteristics We use OLS regression analysis to test how well our model accounts for the role of premiums in the indenture contract. Regressions of the determinants of (log) premium size were estimated for both the nationwide Stamp Tax data and the smaller London

28

in the professions were sons of professionals or gentlemen (Table 4). However, boys from poorer families in agriculture and manufacturing could enter those crafts where premiums were smaller or less frequent. In clothing and footwear, about ten percent of apprentices were labourers' sons. Moreover, premiums

7

Assuming that youths would earn a rising fraction of adult income with age (20% at age 14, 40% at age 15, 60% at age 16, 70% at age 17, 80% at age 18, 90% at age 19, and 20% at age 20 — see Van Zanden, (2009), p. 160), a provincial adult unskilled wage of 1s per day, that youths work 228 days per year (Voth, 2001), and a discount rate of 7.5%, the present value of lost earnings during an apprenticeship, relative to a subsistence income of £5 per annum, was about 26 pounds.

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

345

Table 5 Apprenticeship premium determinants, Stamp Tax. (1) Gentleman father Professional father Primary father Distribution/sales father Service father Labourer father Migrant Migrant X gentleman father Migrant X professional father Migrant X primary father Migrant X distribution/sales father Migrant X service father Migrant X labourer father Father and master in same occupation Migrant X father and master in same occupation Same name master and father Migrant X same name master and father # of apprentices ever taken by master Second apprentice Third apprentice Fourth apprentice Fifth or higher apprentice Ln(av.premium, other indentures) Age 13–15 Age 16–18 Age 19–30 Master citizen Constant Year dummies County dummies Occupation dummies R-square N

(3) Linked to birth or baptism (4) Master with N 1 apprentice (5) London only

(2) c

.71 (.02) .48 (.02) c −.04 (.01) c .46 (.02) c .01 (.02) −.35 (.03) c .08 (.01) c

−.09 (.02) c

c

.65 (.02) .47 (.03) c −.04 (.02) b .44 (.02) c .01 (.02) −.32 (.03) c .04 (.02) b .15 (.03) c .05 (.04) .02 (.03) .06 (.04) a .03 (.04) −.15 (.06) b −.08 (.02) c − .05 (.05)

− .15 (.04) − .15 (.05) − .01 (.08) .01 (.002) b .01 (.01) .07 (.01) c .07 (.01) c .09 (.02) c .14 (.02) c .18 (.03) c .09 (.04) b .11 (.04) c .05 (.04)

.61 (.06) c .55 (.08) c − .05 (.05) .33 (.04) c − .01 (.05) − .20 (.09) b .03 (.05) .16 (.09) a − .09 (.13) − .24 (.09) c .38 (.19) a .07 (.10) − .30 (.18) − .06 (.06) − .28 (.15) a

.26 (.02) c .23 (.03) c − .01 (.02) .18 (.02) c .01 (.02) − .17 (.04) c .01 (.02) .06 (.03) a − .01 (.05) .01 (.03) .05 (.04) a .05 (.04) a .02 (.08) − .04 (.03) .01 (.06)

.61 (.03) c .43 (.05) c − .16 (.05) c .45 (.03) c − .07 (.03) b − .56 (.08) b − .02 (.03) .28 (.04) c .15 (.07) b .16 (.06) c .11 (.05) b .12 (.06) b .04 (.11) − .03 (.04) − .11 (.08)

.12 (.15) − .28 (.24) .01 (.01) a .01 (.03) .07 (.06) − .06 (.09) .11 (.10)

− .15 (.07) b − .05 (.11) − .01 (.01) − .01 (.02) − .01 (.02) − .04 (.03) − .07 (.03) b .72 (.01) c

.06 (.10) − .20 (.14) .01 (.01) .05 (.02) b .10 (.03) c .06 (.05) .01 (.05)

.15 (.06) c .32 (.06) c .15 (.06) b 1.5 (.90) Y Y Y .54 29662

1.5 (.90) a Y Y Y .54 29662

1.9 (1.2) a Y Y Y .52 4263

1.11 (.68) Y Y Y .75 11501

.15 (.02) c 1.2 (0.88) Y N Y .43 14041

Notes: See Table 2 for the sample description. We use all observations with complete apprentice, master, and parent characteristics. Occupation dummies are defined using Wrigley's P.S.T. codes to the third digit. Standard errors are in parentheses. Father craft worker is the excluded parent occupation group. a Significant at the 10% level. b Significant at the 5% level. c Significant at the 1% level.

guild sample.29 While these models contain many similar elements, variable availability and therefore regression specifications are somewhat different in the two samples. For this reason we present results for each source in a separate table. For the nationwide Stamp Tax data (Table 5) all specifications include a full set of guild, occupation and year dummies. Because only positive premiums were recorded, selection into the population is an issue, but we have little capacity to deal with this in a formal manner. We estimated a similar 29 In addition to OLS regressions of premiums size, we have estimated both linear probability and probit models to explain the presence of a premium in the London guild sample. We do not include these results in this article, but they are available in Minns and Wallis (2011), or from the authors on request.

series of regressions for premium size for the London guild sample (Table 6).30 The models in Table 6 have limited information on the master's occupation, although we do use detailed occupations for a sub-sample for which these were recorded. 30

The censored nature of the data is an important issue in estimating regressions of premium size. Various econometric solutions have been proposed to deal with this type of censoring. In the specification used in column 1, we adopt the crude approach of using a left hand side variable equal to ln(premium + £1). Tobit regressions that formally account for the censoring (not reported here) give marginal effects that are similar to the findings in Column 3 of Table 6. We have also estimated the model in Column 3, Table 6 using only the subsample of guild records where positive premiums were recorded. The main results and implications are similar with this variant.

346

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

Table 6 The determinants of London training premiums, 1710–1800. (1) Father citizen Father same company Father deceased Kin apprentice Gentleman father Professional father Primary father Distribution/sales father Service father Labourer father Migrant Migrant X gentleman father Migrant X professional father Migrant X primary father Migrant X distribution/ sales father Migrant X service father Migrant X labourer father Total master apprentices Second apprentice Third apprentice Fourth apprentice Fifth or higher apprentice Father same occupation Bookbinder Bookseller Druggist Grocer Haberdasher Instruments Plasterer Printer Stationer Turner Dyer Tailor Butcher Calenderer Watchmaker Engraver Blacksmith Company Dummies Year Dummies Constant R-square N

(2) a

.19 (.06) − .51 (.06) a .08 (.03) b − .78 (.09) a .98 (.05) a .86 (.07) a − .09 (.05) .45 (.05) a .11 (.05) b − .17 (.08) b .23 (.03) a

−.01 (.004) a .03 (.04) .02 (.04) .12 (.05) b − .03 (.05)

Y Y 2.1 (.21) a .22 11705

(3) a

.21 (.07) − .48 (.06) a .07 (.03) b − .82 (.09) a .74 (.08) a .40 (.10) a − .14 (.10) .34 (.06) a .13 (.06) b − .03 (.11) .09 (.05) c .51 (.11) a .84 (.13) a

.45 (.10) a − .60 (.10) a .13 (.05) a − .92 (.13) a .54 (.10) a .56 (.13) a − .22 (.13) .15 (.07) b .12 (.09) − .08 (.17) .09 (.07) .59 (.15) a .57 (.18) a

.15 (.12) .35 (.10) a

.10 (.17) .61 (.14) a

.13 (.06) b − .03 (.11) − .02 (.01) a .06 (.04) c .11 (.04) a .09 (.05) c .06 (.05)

.12 (.09) − .14 (.28) − .011 (.004) b − .03 (.05) .07 (.06) .09 (.07) − .07 (.06) − .59 (.07) a − .11 (.08) .74 (.09) a 1.1 (.22) a .94 (.16) a 1.7 (.17) a .32 (.12) b − .02 (.25) − .06 (.07) 1.0 (.09) a − .40 (.27) − .37 (.19) c .18 (.21) − .37 (.19) c − .98 (.11) a .63 (.19) a 1.2 (.13) a − .59 (.26) b Y Y .39 (.36) .23 7031

Y Y 1.9 (.21) a .23 11705

Notes: Log premium is equal to ln(Premium + £1), as described in the text. Standard errors are in parentheses. Father craft worker is the excluded parent occupation group. All other occupations is the excluded master occupation group. a Coefficients significant at the 1% level. b Coefficients significant at the 5% level. c Coefficients significant at the 10% level.

The set of explanatory variables we have constructed from the apprenticeship records allow us to test the three alternative explanations for premium size discussed earlier. The first, that premium size served to mediate the likelihood of early departure among apprentices, can be explored using information about the geographic origins and occupational background of those entering training. As noted earlier, attrition among apprentices in the late 17th century was higher for migrant apprentices of relatively prosperous origins (Table 1; Minns and Wallis, 2012). Poor migrants, with fewer outside options, generally stayed. While apprentices' attrition during contract is unobserved in the two data sets for which we have premiums, the information they contain on apprentices' origins allows us to explore whether premiums were higher for those whose profile suggests that they were likely to depart early. The Stamp Tax and Company records also allow us to construct several measures of the potential productivity of apprentices and their masters, such as the age of the apprentice, the occupation of their father and the master's prior experience with apprentices. If the price of training was responsive to differentials in expected productivity and training quality, as indicated in the second explanation, these characteristics should account for some of the heterogeneity of premiums within occupations (as observed in Fig. 3), and should even retain some significance when comparing different apprentices who signed terms with the same master. Finally, if the conjecture that premium size reflected the rents available because guilds created an artificial scarcity of training is valid then prices should vary widely between guilds and non-guild members, and be highest for guilds that retained most power to influence the training market. 6. Explanation 1: premiums and risk of departure The regressions offer support for the hypothesis that premiums offered compensation for limited information and higher risks. Distance offers one proxy for information: masters were likely to know less about migrant apprentices. In both the London and Stamp Tax samples, we find that migrant apprentices (not from the county in which the master was located) paid significantly larger premiums (Table 5, column 1; Table 6, column 1).31 The London sample shows only a small We also estimated this model with a “neighbouring counties” variable for counties bordering London or Middlesex (Berkshire, Buckinghamshire, Essex, Hertfordshire, Kent, and Surrey), with similar results. 31

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

difference between remote and local apprentices in the likelihood that a premium was levied: better information might produce discounts, but few waivers. Family connections, which carried information but also social obligations, had the biggest impact on the likelihood of avoiding a premium. For London companies we have direct evidence in the form of kin ties, while the Stamp Tax shows similar (though smaller) effects among apprentices who share a surname with their master. As this underlines, information is often embedded in social networks that can themselves reduce risk and which might also produce alternative reasons for discounts through mutual obligations.32 Do premiums reflect the probability that apprentices would depart early? When we include interaction terms between being a migrant and parental occupation, we find that the interactions are positive, statistically significant, and large in many cases (relative to the excluded group of sons of craft workers). The interaction coefficients are larger and of greater statistical significance when we restrict attention to London apprentices (Table 5, column 2; Table 6, column 2), where we know that attrition was at high levels. Premiums were largest for those likely to have the best outside options: the sons of gentlemen, traders or professionals. Conversely, migrant apprentices from low-status backgrounds – labourers' sons – paid no more, perhaps less, than migrant craft workers' sons, all else equal. These patterns closely match the probability of departure for these different groups we observed in work on attrition in the 1690s (Minns and Wallis, 2012). Apprentices appear to have paid extra to compensate masters for the increased probability of early departure: for gentlemen's sons, premiums were as much as 20% higher. Interactions of migrant status with shared occupation and shared surname yield insignificant coefficients in our baseline model and most other specifications.

7. Explanation 2: premiums and productivity The regressions also show how premiums were used to respond to the heterogeneity of apprentices and 32 We focus here on the effects of information and social networks on the price of training among those who served an apprenticeship. A related issue is whether information transmitted through such networks provides a net social benefit, or simply reserved a greater share of opportunities for insiders (Ogilvie, 2004). There is some evidence for the second process in English apprenticeship and guild systems where guild fees and entry rules differentiated between males and females and sons of masters and those of others. However, we would expect such effects to be a minor factor in setting premium levels, as these are negotiated privately whereas most effective barriers to entry are set collectively in order to minimise free-riding.

347

masters in terms of differences in anticipated productivity (v), subsistence expectations (w), and institutional rents (B). The analysis establishes a clear link between productivity and premiums. Apprentices with relevant prior experience through exposure to the occupation received a discount: those whose fathers' occupations matched their master's paid premiums that were 8 to 9% lower in our baseline sample (Table 5, columns 1 and 2). Age offers a more surprising result. As physical maturity would affect apprentices' productivity, one would expect younger apprentices to pay more. Instead, premiums peaked around the age of 17 (Table 5, column 3), even in craft occupations, before falling for the small group of older apprentices who likely had prior skills.33Very young apprentices may have been serving longer terms.34It is possible that the rising age-premium profile reflects the increased risk of attrition among older apprentices who were more likely to succumb to the temptations of marriage and wage labour. The results also show that, even within occupations, premiums reflected apprentices' backgrounds. The relationship between parental occupation and premium appears broadly similar in both samples. Premiums were larger, and more common, for the children of more prosperous fathers, such as gentlemen, professionals and those in commerce and trade, while the children of primary-sector fathers, husbandmen and labourers paid less. For London, this was not driven by occupational variation within guilds: when we use master's occupation (Table 6, column 3), parental background variables remain large and statistically significant.35Due to the large numbers of buyers and sellers in the British market for training it seems unlikely that masters were able to use price discrimination to charge the wealthy more. More plausible is that wealthier families paid more for better conditions: they supplied a larger P so their children enjoyed a greater w while indentured. This explanation had prominent advocates in the period: Daniel Defoe (1726, p. 12, p. 148) thought that high premiums were because parents' ‘unreasonable fondness and partiality’ towards their children led them to pay more in order save their offspring from menial work. 33

If we estimate this regression for the construction sector alone, premiums peak at age 18, controlling for all other characteristics. 34 See Reith (2007), and De Munck (2007). Galenson (1981) finds that younger indentured servants had longer terms, all else equal. 35 The coefficients on gentleman, distribution/sales, and professional father are larger in this sub-sample (column 5) than in the broader sample (columns 3 and 4), with or without controls for master occupation.

348

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

Limited information about likely match quality affected apprentices as well as masters. The results show that apprentices valued demonstrable training quality. We are able to add information about masters' training careers to the regression models in the form of the total number of apprentices trained, and the rank in the master's training profile of each individual apprentice. Compelling evidence that masters' experience of training mattered comes through the effects of apprentice rank. The Stamp Tax results show that masters were able to charge higher premiums for later apprentices, with the fourth apprentice paying about 10 to 20% more than the first (Table 5, column 1 and 2).36In the London Stamp Tax sub-sample (column 3), the effects of rank are somewhat smaller, but broadly in line with what we find in the sample of London companies (Table 6, column 1 and 2). We can extend this approach to get some sense of the effect of other, unobservable characteristics of masters. We include the log of the average premium each master charged 37in other indentures, for masters who trained at least two identifiable apprentices (Table 5, column 4). Unsurprisingly, an apprentice's premium was highly correlated to that paid by other apprentices to the same master. This is perhaps the clearest indication of how master-specific attributes determined premiums even within well-defined occupations. In this model, most apprentice characteristics and master experience still affect premiums, with positive and significant coefficients on the usual parental background variables. That premiums varied with these characteristics conditional on the magnitude of the premium their master would typically receive offers compelling evidence that the apprenticeship market was highly responsive to differences in information, expectation and experience.38 8. Explanation 3: premiums and scarcity As we suggested earlier, formal institutions did not secure apprenticeship contracts. However, many apprenticeships who completed terms would expect to 36

Unfortunately we do not know how successful previous indentures may have been. The measure of number of apprentices and their rank that we use will suffer from downward bias, as we miss any apprentices for whom no premium was paid or recorded. 37 For apprentice j, the mean premium charged to other apprentices ð∑n P Þ−P trained by the same master is equal to P others;j ¼ i¼1n−1i j : 38 We have also experimented with a fixed effect version of this regression, with controls for each individual master. This specification, which effectively adds several thousand dummy variables to the model, yields similar point estimates on the key variables related to background and apprentice sequence, but with larger standard errors.

capture institutional rents. Elsewhere we have shown that the tranche of apprentices who did complete contracts sought to comply with formal regulations, even if this meant returning to their master after an absence (Minns and Wallis, 2012). Here, the large price differences between guilds and occupations reported in Tables 2 and 3 retain their magnitude and statistical significance after controlling for master and apprentice characteristics within each indenture. This finding is illustrated in detail in Table 6, column 3, where we report the wide range in premium prices between selected occupations. It is likely that these coefficients reflect both the effects of limited training places for high-demand occupations, and differences in the quality and quantity of training provided to apprentices in certain occupations. A more focused test of the idea that the price of apprenticeship reflected the possibility to rent-seek after training is provided in regressions indicating statistically and economically significant price differences between masters who were citizens of London (and therefore guild members) and those who were not. Stamp Tax apprentices paid about fifteen percent more to train with a citizen (Table 5, column 5). If we compare masters with specified occupations in the guild sample to London masters who were not citizens in the Stamp Tax data we find that among grocers, booksellers, stationers and printers, citizens received significantly larger premiums: a 124% difference for grocers (£101 versus £45), 96% for booksellers (£90 versus £46), 139% for stationers (£79 versus £33), and 20% for printers (£21 versus £17.5). For plasterers, however, premiums were much closer: 7.7 pounds with a citizen and 7.0 with a non-citizen.39 The variation between occupations above fits well with the historiography of these different trades. Guild membership as a plasterer meant little, with any guild rents permanently diminished following the sector's de-regulation to encourage rebuilding after the Great Fire of 1666. The Grocers' Company showed little interest in defending monopolies in this period, but it did possess prestige and the possibility of useful networks. The Stationers' Company (which contained booksellers, stationers and printers) was an unusually strong and homogenous guild, and asserted its authority over the printing and book trades throughout this period. This pattern confirms that apprentices training 39

For London citizens, company not occupation was recorded. Thus, here we compare those company records that separately record occupation to non-citizens in the Stamp Tax for whom only an occupation is available.

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

in guilds with significant market power were willing to pay larger premiums in anticipation of post-completion rents and further economic privilege. 9. Conclusion Apprenticeship was the main source of non-agricultural training in pre-industrial England. The scale and continuity of the practice suggest that it was a success, despite small firm size, limited information flows about prospective training partners, and low completion rates. Masters and apprentices were heterogeneous, but England's rigid labour laws gave them little room to acknowledge this through their contracts. Our analysis shows that training premiums were used to address the core tensions in securing pre-industrial training contracts. Premiums compensated apprentices and masters for the anticipated risks of default and limited information about training quality, supplying a crucial degree of freedom that allowed masters and apprentices to reach long-term indenture agreements. Premiums also reflected occupational returns and residual corporate privileges. England's reliance on premiums to create contracts increased the price of entrance into apprenticeship. This had consequences for access to training among the credit constrained, but premiums were typically small relative to opportunity costs and the capital required for a business. Apprenticeship premiums were not ubiquitous, and even in occupations where they were charged frequently they were modest relative to expected incomes. Premiums do not represent the only potential solution to the hold-up problem in a pre-industrial training market. For example, labour laws could have been altered to allow shorter, more flexible terms of indenture, as in other markets for bound labour in the period. However, proposals to change labour law faced strong opposition (Rule, 1981; Berlin, 2008), and neither the state, which taxed premiums, nor masters who received them had strong incentives to change the system. Understanding how training was secured has implications for the ongoing debate regarding the role of formal institutions in generating economic growth. England's precocity in human capital formation has been identified as a key engine of early economic growth, but the restrictive framework surrounding apprenticeship indentures in pre-industrial England gave masters and apprentices few degrees of freedom with which to form mutually agreeable contracts. Evidence of high rates of attrition suggests that some aspects of these contracts were in any case unenforceable, and instead of enforcement, English legal institutions offered relatively easy dissolution of contracts (Wallis, 2012). In these circumstances,

349

premiums were a private-order arrangement that allowed training contracts to be adjusted to individual circumstances, despite high levels of uncertainty, limited information flows, and the rigidities created by English labour law. While many economic historians have argued that English pre-industrial institutions were one of the foundations upon which modern economic growth was built, it is striking that a private-order, informal mechanism was a necessary complement to formal rules in this crucial area of human capital formation. References Acemoglu, D., Pischke, J.-S., 1999. Beyond Becker: training in imperfect labour markets. The Economic Journal 109 (1), 112–142. Allen, R.C., 2001. The great divergence in European wages and prices from the middle ages to the first world war. Explorations in Economic History 38 (4), 411–447. Becker, G., 1964. Human Capital. National Bureau of Economic Research, New York. Ben-Amos, I.K., 1991. Failure to become freemen: urban apprentices in early modern England. Social History 16 (2), 155–172. Ben-Amos, I.K., 1994. Adolescence and Youth in Early Modern England. Yale University Press, New Haven. Berlin, M., 2008. Guilds in decline? London livery companies and the rise of a liberal economy, 1600–1800. In: Epstein, S.R., Prak, M. (Eds.), Guilds, Innovation, and the European Economy. Cambridge University Press, Cambridge, pp. 1400–1800. Brooks, C.W., 1994. Apprenticeship, social mobility, and the middling sort, 1550–1800. In: Barry, J., Brooks, C.W. (Eds.), The Middling Sort of People: Culture, Society and Politics in England, 1550–1800. Macmillan, Basingstoke, pp. 52–83. Campbell, R., 1747. The London Tradesman, Being a Compendious View of All the Trades, Professions, Arts, Both Liberal and Mechanic, Now Practised in the Cities of London and Westminster. T. Gardner, London. Centre for Metropolitan History, 2012. Records of London's Livery Companies Online. www.londonroll.org (last accessed 6 February 2013). Clark, G., 2005. The condition of the working class in England, 1209–2004. Journal of Political Economy 113 (6), 1307–1344. Davies, M.G., 1956. The Enforcement of English Apprenticeship, 1563–1642: A Study in Applied Mercantilism, 1563–1642. Harvard University Press, Cambridge. Defoe, D., 1726. The Complete English Tradesman. Charles Rivington, London. De Munck, B., 2007. Technologies of Learning: Apprenticeship in Antwerp Guilds from the 15th Century to the End of the Ancien Regime. Brepols, Turnhout. de Vries, J., 1984. European Urbanization, 1500–1800. Methuen, London. Earle, P., 1989. The Making of the English Middle Class: Business, Society, and Family Life in London, 1660–1730. University of California Press, Berkeley. Epstein, S.R., 1998. Craft guilds, apprenticeship, and technological change in preindustrial Europe. The Journal of Economic History 58 (3), 684–713. Galenson, D.G., 1981. Market evaluation of human capital: the case of indentured servitude. Journal of Political Economy 89 (3), 446–467.

350

C. Minns, P. Wallis / Explorations in Economic History 50 (2013) 335–350

Go, S., Lindert, P.H., 2010. The uneven rise of American public schools to 1850. The Journal of Economic History 70 (1), 1–26. Grassby, R., 1995. The Business Community of Seventeenth-century England. Cambridge University Press, Cambridge. Grubb, F., 1997. Does bound labor have to be coerced labor? The case of colonial immigrant servitude versus craft apprenticeship and life-cycle servitude-in-husbandry. Itinerario 21 (1), 28–51. Hamilton, G., 1996. The market for montreal apprentices: contract length and information. Explorations in Economic History 33 (4), 496–523. Humphries, J., 2003. English apprenticeship: a neglected factor in the industrial revolution. In: David, P.A., Thomas, M. (Eds.), The Economic Future in Historical Perspective. Oxford University Press, Oxford, pp. 73–102. Humphries, J., 2009. Through the Mill: Child Labour in the British Industrial Revolution. Cambridge University Press, Cambridge. Kaplan, S.L., 1993. L'Apprentisage au XVIIIe Siècle: Le Cas de Paris. Revue d'Histoire Moderne et Contemporaine 40 (3), 436–479. Kussmaul, A., 1981. Servants in Husbandry in Early Modern England. Cambridge University Press, Cambridge. Leunig, T., Minns, C., Wallis, P.H., 2011. Networks in the premodern economy: the market for London apprenticeships, 1600–1749. The Journal of Economic History 71 (2), 413–443. Lindert, P.H., 2004. Growing Public: Volume 1, The Story: Social Spending and Economic Growth Since the Eighteenth Century. Cambridge University Press, Cambridge. McKenzie, D.F., 1978. Stationers' Company Apprentices 1701–1800. Oxford Bibliographical Society, Oxford. Minns, C., Wallis, P.H., 2009. Apprenticeship and skill in eighteenth century England: the decline of apprenticeship in London. Paper Presented at the 2009 World Economic History Congress. Minns, C., Wallis, P.H., 2011. Why did (pre-industrial) firms train? Premiums and apprenticeship contracts in 18th century England. LSE Economic History Working paper 155. Minns, C., Wallis, P.H., 2012. Rules and reality: quantifying the practice of apprenticeship in early modern England. The Economic History Review 65 (2), 556–579. Mokyr, Joel, 2009. The Enlightened Economy: An Economic History of Britain, 1700–1850. Yale University Press, New Haven. More, C., 1980. Skill and the English Working Class, 1870–1914. Croom Helm, London. Muldrew, C., 1998. The Economy of Obligation: The Culture of Credit and Social Relations in Early Modern England. Palgrave Macmillan, Basingstoke.

Ogilvie, S.G., 1997. State Corporatism and Proto-Industry: The Württemberg Black Forest, 1580–1797. Cambridge University Press, Cambridge. Ogilvie, S.G., 2004. Guilds, efficiency, and social capital: evidence from German proto-industry. The Economic History Review 57 (2), 286–333. Reith, R., 2007. Apprentices in the German and Austrian crafts in early modern times: apprentices as wage earners? In: DeMunck, B., Kaplan, S.L., Soly, H. (Eds.), Learning on the Shop Floor: Historical Perspectives on Apprenticeship. Berghahn, New York, pp. 179–203. Rule, J., 1981. Experience of Labour in Eighteenth-Century Industry. Croom Helm, London. Shaw-Taylor, L., 2010. The Occupational Structure of England c. 1750–1871: A Preliminary Report. Unpublished Working Paper, Cambridge University. Smits, W., Stromback, T., 2000. The Economics of the Apprenticeship System. E. Elgar, Cheltenham. Snell, K.D.M., 1996. The apprenticeship system in British history: the fragmentation of a cultural institution. History of Education 25 (4), 303–322. van der Beek, K., 2010. Technology-Skill Complementarity on the Eve of the Industrial Revolution: New Evidence from England (1710–1772). Unpublished manuscript, Ben-Gurion University. van Zanden, J.L., 2009. The skill premium and the ‘great divergence’. European Review of Economic History 13 (1), 121–153. Voth, H.J., 2001. The longest years: new estimates of labor input in England, 1760–1830. The Journal of Economic History 61 (4), 1065–1082. Walker, M.J., 1986. The Extent of Guild Control of Trades in England C. 1660–1820: A Study Based on a Sample of Provincial Towns and London Companies. Cambridge Universty PhD Thesis. Wallis, P.H., 2008. Apprenticeship and training in premodern England. The Journal of Economic History 68 (3), 832–861. Wallis, P.H., 2012. Labour, law and training in early modern London: apprenticeship and the city's institutions. Journal of British Studies 51 (4), 791–819. Wallis, P.H., Webb, C., Minns, C., 2010. Leaving home and entering service: the age of apprenticeship in early modern London. Continuity and Change 25 (3), 377–404. Wrigley, E.A., 2004. Poverty, Progress, and Population. Cambridge University Press, Cambridge.