Introduction
Female servants constituted the largest single occupational group in eighteenth- and nineteenth-century England and Wales.Footnote 1 Jane Humphries estimates that service represented the entry into the labour market for twenty-eight per cent of girls during industrialization.Footnote 2 Scholars such as Jacob Field, Edward Higgs, Ann Kussmaul, Charmian Mansell, and Jane Whittle have contributed substantially to our understanding of service. However, there remain gaps in our knowledge regarding the terms and conditions of their employment, particularly in relation to the functioning of the labour market for servants.Footnote 3
Wages are central to understanding historical economic development and social realities, yet wage formation remains inadequately understood. Research into the origins of the Industrial Revolution and early modern labour markets has yielded contrasting insights into wage levels and formation. For example, Robert Allen argues that high wages drove investment in labour-saving technologies, while Judy Stephenson contends that the male construction wage series underpinning Allen's argument were actually payments to contractors.Footnote 4 The recognition of labour and labour markets as key to industrialization underscores the importance of reassessing the labour market of the largest occupational group – servants – using new sources and theoretical frameworks.
Recent debates on early modern labour markets have focused on the extent of their competitiveness.Footnote 5 Historians have extensively discussed institutions, such as guilds, which impeded the efficient functioning of markets, but other restrictive practices conceivably also permeated these markets, inhibiting their operation.Footnote 6 Contemporaries were indeed aware of anti-competitive practices. Adam Smith, for instance, remarked that “masters are always and everywhere in a sort of tacit, but constant and uniform combination, not to raise the wages of labour above their actual rate”.Footnote 7 While historians have long acknowledged power imbalances between employers and workers, economists, in recent decades, have moved away from competitive conceptions of labour markets and have begun to account for imbalances in market power between employers and workers.Footnote 8
In contrast to a perfectly competitive market, where the labour supply is perfectly elastic and workers receive their marginal product, in static monopsonistic markets – for example, a one-company town – the employer has the power to set wages.Footnote 9 Based on recent empirical findings of employer market power in diverse settings, Alan Manning has incorporated these insights into the concept of “dynamic oligopsony”. He posits that, in labour markets with multiple employers, collusion among employers, the depth of the labour market, and the costs for workers associated with job-to-job transitions grant employers a degree of discretion in setting wages and conditions.Footnote 10 This theoretical framework offers historians a valuable lens through which to analyse labour markets and wage formation. Yet, these new frameworks have not been applied to servants’ wages. The literature frequently assumes, often implicitly, that the market for servants was competitive.Footnote 11 This assumption has far-reaching implications, since only by understanding the gender-disaggregated processes of wage formation can we, for instance, begin to unravel historical gender pay gaps.Footnote 12
This paper examines the micro-level determinants and nominal wage patterns of female servants’ wages in England and Wales between 1780 and 1834 to assess the competitiveness of the labour market. The study utilizes a previously unused dataset of over 1,400 distinct employment relationships recorded in the single women settlement affidavits of the General Lying-In Hospital (GLIH), a maternity hospital situated on the Thames's south bank in the rapidly urbanizing parish of Lambeth, Surrey. It addresses two questions: (1) What was the influence of servants’ individual characteristics, the nature of the employment relationship, and the wider socio-economic context on servants’ wages? (2) How did the competitiveness of the servant labour market change? In what follows, Section 1 introduces the sources and evaluates regional wage patterns against existing series. Section 2 presents a regression analysis to estimate the micro-level determinants of servants’ wages. Section 3 examines trends in employer market power and discusses its sources using qualitative evidence, such as servants’ autobiographies. Finally, some brief observations conclude this paper.
The Data and Regional Wage Patterns
The GLIH served both single and married working-class women.Footnote 13 To prevent the parish of St Mary Lambeth from being charged with the maintenance of newborns, the law (Lying-in Hospitals Act 1773, 13 Geo. 3. c. 82) required that expectant mothers’ parochial settlement be documented at the hospital prior to delivery. A Justice of the Peace oversaw this examination approximately two months before the expected birth.Footnote 14 Women could attain parochial settlement by birth, apprenticeship, yearly hired service, renting a dwelling valued at more than £10 per annum, or marriage. Of the single women admitted, 1,435 had gained settlement through annual live-in service, and their wages were recorded for the period between 1780 and 1834.Footnote 15 These expectant mothers had worked as servants across England and Wales (Appendix 1). Both London and the GLIH attracted women from throughout England, with at least one wage recorded for every county except Cumberland, Rutland, and Westmorland.Footnote 16 Coverage for Wales was patchier, limited to central and northern regions.
Of the female servants’ wages recorded in the single women's settlement affidavits, 60.24 per cent came from London and Middlesex, 20.54 per cent from the Home Counties, 8.7 per cent from the Southeast, 4.6 per cent from the Southwest, 3 per cent from the West Midlands, 2.23 per cent from the North, and 0.7 per cent from Wales. The settlement affidavits recorded the date of the examination, the parish and county of settlement, the duration of the servant's employment, the employer's name (and title), the nominal yearly wages paid, and the applicant's signature or mark.Footnote 17 A unique strength of this study is that each observation constitutes a distinct employment relationship.
The GLIH settlement affidavits record a wide range of servants’ economic entitlements, from room and board only to as much as £63 in addition to room and board. A histogram of wages is shown in Figure 1, and the clustering of wages is discussed in Section 3. A typical example is Emma McPheerson, who, when examined in 1809, had served in Tottenham for more than three years, receiving eight guineas per annum, and signed her affidavit.Footnote 18 In about a dozen cases, women received no wages, working solely in exchange for room and board, which at times was supplemented by occasional monetary payments.Footnote 19 Mary Ann Massey, for example, worked as a servant in Lambeth, Surrey, and received “occasional pocket money” for her services.Footnote 20
Figure 1. Histogram of distribution of nominal wages. Note: The chart excludes four outliers.
Sources: LMA/H01/GLI/B/09/001; LMA/H01/GLI/B/10/001–003; LMA/P85/MRY/231–238.
Although all servants received in-kind compensation, those employed in well-heeled households might have received more or better-quality perquisites, though the sources remain silent on their value. A recent strand of literature highlights the importance of in-kind payments within labour contracts and, importantly, identifies them as a source of employer market power.Footnote 21
The affidavits do not distinguish between domestic and farm service, but strictly delineating these two categories is problematic. Both types of service involved the servant residing with her employer, and her time was entirely at the employer's disposal. By the mid-eighteenth century, law manuals, as noted by Carolyn Steedman, defined service according to these characteristics rather than by the tasks performed.Footnote 22 Integrated into her employer's household, a servant typically carried out whatever tasks required attention; in rural areas, domestic servants often assisted with agricultural work.Footnote 23
The risks of conflating farm and domestic service are greatest when comparing wage sources from aristocratic and working-class households. However, in this study, the issue is minimized, as most of the servants examined at the GLIH worked in small establishments. Only 11.5 per cent were employed by individuals with a status prefix (e.g. Rev, Dr, Lord), and just 6.4 per cent were employed by farmers.Footnote 24 Appendix 4 details the occupations of the employers of servants examined at the GLIH. It is likely that most of these women worked in single-servant households as generalists.
Charmian Mansell observes that, while the tasks performed by servants varied across households along the social spectrum, her data also suggests that most servants, regardless of their employer's social status, undertook a mixture of duties.Footnote 25 Moreover, autobiographical accounts from servants reveal that even in elite households, they were often deployed flexibly. For example, Mary Ann Ashford, during her three positions in wealthy households in late eighteenth-century London, chopped firewood, pumped water out of a flooded cellar, provided childcare, and cared for her employer's birds, cats, dogs, and monkey.Footnote 26
Certainly, the servants examined at the GLIH are not a random sample of all servants. First, they were examined in the context of being pregnant and receiving charity. While it is difficult to meaningfully assess the extent to which pregnancy was aberrant, it is possible to speculate about the selectivity involved, given that these women were recipients of charity. As eloquently described by Mary Ann Ashford, access to charity was governed by a complex web of social interactions, dependent on the performance of respectability and the maintenance of patronage networks.Footnote 27 Therefore, the single women who gave birth at the GLIH certainly had more social resources to draw upon than their counterparts in the workhouse.
Second, many of the women examined at the GLIH had migrated to London, raising concerns about possible migration bias. However, the arguments presented in this paper regarding employer power suggest that the very act of migration reflects a level of initiative. This, in turn, implies that these women may have been less susceptible to employer control.
Third, although the ages of the women examined are not recorded, the maternity delivery books (1826–1855) indicate that the age range of both single and married women admitted was from around eighteen to their mid-forties.Footnote 28 Due to the large number of women passing through the hospital, they had varying lengths of service, including some who were long-term servants (Appendix 7).
Finally, to address the potential issue of migration bias, the analysis closely replicated the ordinary least squares model used by Jacob Field to examine wage variation and assess selection bias (Appendix 5).Footnote 29 The settlement affidavits reflect the same pattern of regional wage variation that Field identifies.Footnote 30 Although an imperfect test, the broad alignment of regional coefficients suggests that the wages recorded at the GLIH do not constitute an outlying sub-sample.
Determinants of Servants’ Wages
This section presents a spatial model and a set of interaction terms to test the relationships between servants’ real wages, literacy, tenure, alternative employment opportunities, welfare expenditure, and wealth inequality. As depicted in Table 1, the analysis of the wage data from the GLIH affidavits employed a generalized spatial two-stage least squares model with a spatially lagged dependent variable and a set of interaction terms to estimate the correlates of the logarithm of servants’ real wages.Footnote 31 A spatial regression model with a spatially lagged dependent variable accounts for spatial dependencies in the dependent variable. In this context, spatial dependencies were likely, as the wages paid to one servant could influence the wages paid to another through market competition or labour mobility.Footnote 32 The key advantage of using a spatial lag model is that it estimates whether wages were locally correlated, thereby providing insights into how localized labour markets were.
Table 1. Generalized spatial two-stage least squares model with a spatially lagged dependent variable with interaction terms.
Legend: Standard errors in parentheses; *** p < 0.01; ** p < 0.05; * p < 0.1.
The regression model ran iterations on eight specifications (columns 1–8 in the table). A dummy variable, “Literacy”, served as a proxy for the servants’ human capital, based on whether a woman signed or marked her affidavit (literate being assigned the value of 1 and illiterate the value of 0). “Tenure” was measured as a dummy variable, assigned the value of 1 if the servant served for more than one year and 0 if one year or less. “Status” indicated whether the servant's employer had a status prefix before their name, with the value of 1 assigning elite households and 0 non-elite households.Footnote 33 Given the heterogeneity of the term “service”, the analysis explored the interactions between literacy, tenure, and employer social status, with illiterate single-year servants in non-elite households functioning as the reference group (“Ref.”) in all specifications; that is, the baseline against all other groups were compared. This three-way interaction of terms allowed for a more nuanced understanding of the combined effects of literacy, tenure, and employer status on servants’ wages, enabling an analysis of how, for example, the interaction between literacy and tenure varied depending on whether the servant worked for an elite or non-elite employer.
The study estimated changes in servants’ real wages using time dummies (e.g. 1780–1799). The variable “Distance from London” represented the logarithmic distance, in kilometres, from the servant's parish to London, the largest and most diversified urban economy in England and Wales at the time. “Distance from Manchester” did the same but for Manchester. Proximity to deeper labour markets, such as London's diversified economy and Lancashire's textile industries, should be associated with higher wages. Migration involved substantial upfront costs, but cottage industries may have offered local women an alternative to service. To capture the local dimension of labour market depth, the variable “Cottage industry” reflected the share of parishes in the county of settlement where female employment opportunities existed in lacemaking, knitting, or spinning. Additionally, the study employed a broader measure of local labour market depth, using the variable “Alternative employment”, which substituted cottage industries with the share of parishes in the county of settlement where work for women was recorded.Footnote 34 The analysis estimated the relationship between wealth inequality and wages by controlling for the log of “Wealth per capitalist” in the county of settlement.Footnote 35 It also accounted for the generosity of poor relief by controlling for decadal levels of log “Relief per capita”.Footnote 36 As a robustness check, I re-ran the models excluding observations from London and Middlesex (columns 5–8).Footnote 37
The status of household (elite or non-elite) in which a servant worked proved to be a more significant predictor of wage than either her literacy (literate or illiterate) or the duration of her tenure (single-year or multi-year). While this was likely due to unobserved characteristics of the servant and her productivity, a comparison of the interaction terms is revealing. As can be seen in Table 1, an illiterate single-year servant in an elite household could expect to be paid between 21.9 per cent and 35.8 per cent more than an illiterate multi-year servant in a non-elite household. The wages of single-year illiterate women in elite households did not differ significantly from those of single-year literate servants in non-elite households,Footnote 38 nor were there pay differentials between illiterate multi-year servants in elite households and literate multi-year servants in non-elite households.Footnote 39 Within elite households, there was little pay stratification according to literacy. Literate multi-year servants wages were insignificantly different from illiterate single-year servant's wages.Footnote 40
Furthermore, there were no significant differences between the wages of literate single-year and multi-year servants,Footnote 41 between literate single-year servants and illiterate multi-year servants,Footnote 42 between literate single-year and illiterate single-year servants,Footnote 43 or between literate and illiterate multi-year servants.Footnote 44 Thus, the only significant wage differences were between literate multi-year servants and illiterate single-year servants, arguably representing the top and bottom of the servant hierarchy.
Since literacy directly impacted a servant's productivity, either employers captured the productivity gains or servants were compensated in-kind. First, reading and writing were useful when conducting commercial tasks on behalf of an employer or reading instruction manuals.Footnote 45 Second, for servants responsible for children, the ability to read to them was a valued skill, keeping them entertained, contributing to their education, and imparting moral values. Finally, literacy functioned as a marker of social differentiation; its absence signified a lack of respectability, with some household manuals advising that illiterate women should not be hired as servants.Footnote 46
Perhaps in elite households, where servants specialized in specific tasks, literacy – a form of general human capital – was less valuable than respectability or the skills acquired over the course of employment. While Field hypothesizes that education and skill were the most important factors in determining wages, this analysis suggests that, within elite households, literacy did not attract a premium.Footnote 47 By contrast, in non-elite households, literacy commanded a wage premium: in such settings, literate servants earned between 7.7 and 16 per cent more than illiterate servants, regardless of tenure. Non-elite households likely placed greater emphasis on a servant's productive capabilities, whereas in elite households, servants formed part of the employer's conspicuous consumption, with greater focus on non-observable factors such as respectability. In sum, the returns to literacy were contingent upon the type of household (elite or non-elite) in which servants were employed.
Wage progression associated with tenure appears to have been the exception rather than the rule. In non-elite households, multi-year servants earned no more than single-year servants, regardless of literacy level.Footnote 48 Equally, in elite households, there were no wage returns to tenure for servants.Footnote 49 These findings align with those of Jane Whittle, who documented male and female servants’ wages over the course of their tenure in a small number of elite households in early modern England. She found that there were limited wage gains for men and none for women, which she attributed to the shorter duration of service performed by women.Footnote 50 The lack of tenure-related pay progression within an employment relationship may be explained by several factors: the absence of productivity increases over tenure, the possibility that productivity gains were compensated only through in-kind benefits, or the idea that employers captured these gains. However, the hypothesis that there were no productivity increases over the course of the employment relationship has been rejected in the historiography. Mansell contends that “service taught young women economic management and business acumen”, while Whittle suggests that the skills women acquired during their employment in service formed the basis of gendered human capital complementarity in marriage.Footnote 51
Moreover, given that servants catered to the individual preferences of their employers, it is theoretically plausible that a servant's productivity increased as she became more familiar with her employer's preferences. Servants may have been compensated for tenure through in-kind benefits or non-material improvements to their labour contracts. Evidence from the early modern construction sector, for example, shows that longer tenure was rewarded with a more consistent supply of work.Footnote 52 In elite households, in particular, the scope for material non-wage improvements in compensation was likely the greatest. However, the weak independent relationship between tenure and money wages is consistent with contemporaneous complaints that servants often sought higher-paying positions after acquiring skills, despite the investments made in their training by their current employers. One household manual lamented:
The poor are ungrateful, and that after all the pains and trouble which may have been taken in making them good servants, it often happens, that instead of testifying a proper sense of the obligation, they become restless, and desirous of leaving those who have had all the trouble of qualifying them for better places and higher wages.Footnote 53
In a labour market, characterized by efficient employer–employee matching, wages would be expected to increase as workers gain experience and skills, reflecting their increased productivity. However, the results here suggest that servants had to leave their employers to secure higher wages, meaning that employers could resist wage demands, even as productivity rose. This would have imposed significant costs on servants, as securing wages commensurate with their skills required changing employers, which came with the associated risks and costs of finding new employment (see Section 3). The lack of tenure-related wage progression could also be attributed to the fact that employers exploited the costs associated with job changes, enabling them to capture productivity gains without having to raise wages. Contemporary labour market research shows that an attenuated relationship between tenure and wages is indicative of employer power, as workers’ compensation should rise in line with their increasing productivity if they are gaining skills during their employment.Footnote 54 In this context, the findings suggest that the relationship between servant and employer was shaped by market power imbalances, which allowed employers to capture, at least in part, the productivity gains servants made over the course of their employment.
The proxy variables measuring labour market competitiveness presented a nuanced picture. Distance from London (thirteenth row) emerged as a significant negative predictor of servants’ wages, though the results were not robust when observations from London were included (columns 3–4). However, when excluding London (and Middlesex), a one per cent increase in the logarithmic distance from London was associated with a decrease in servants’ wages of between 4.5 and 6.2 per cent (columns 5–8). This suggests that the London labour market was relatively competitive, helping to explain why so many young women migrated to the metropolis. The presence of the royal court and aristocracy, alongside a large commercial merchant class, created a concentration of high-paying service positions, giving the service labour market depth.Footnote 55 Additionally, London's diversified economy offered employment options outside of service,Footnote 56 likely driving up wages in the service sector and improving servants’ bargaining positions. More broadly, research has shown that deeper labour markets exhibit lower rates of monopsony.Footnote 57 George Boyer, for instance, found that distance from London was negatively associated with the earnings of male agricultural labourers.Footnote 58 Taken together with Boyer's findings, the results here suggest that London's diversified and dense labour market boosted wages for workers across both gender and occupation.
While London's large and varied economy drove up wages in its vicinity, other indicators of labour market competitiveness suggest that the service labour market outside of London was segmented and localized. Distance from Manchester (fourteenth row of Table 1) was insignificant across all specifications, indicating that women could not easily transition from service to the emerging textile industry in Lancashire. Similarly, the presence of cottage industries or broader alternative employment opportunities was not associated with higher wages for servants (except column 6), suggesting that labour markets were shallow and segmented. These conclusions were reinforced by the insignificant results of the lagged dependent variable (the log of the real wage), which show that servants’ wages were not significantly influenced by the wages of other servants. Wage-setting appears to have been highly localized, further evidence of employers’ substantial wage-setting power.
Several reasons explain why the service labour market was segmented from alternative female employment opportunities. First, the compensation package in service was distinct from that in other industries, with room and board forming a vital component of the former. Autobiographical evidence suggests that the prospect of yearlong contracts, and the associated security of room and board, was a major factor in the appeal of service.Footnote 59 For instance, Rose Allen, after the death of her father, took up sewing to support herself and her ailing mother. The costs of room and board became such a burden for the bereaved family that they decided Allen should enter service to reduce household expenses.Footnote 60 Similarly, after being orphaned in early adolescence, Mary Ann Ashford sought advice from her kin and her parents’ friends on whether she should pursue an indoor apprenticeship or domestic service. Her father's friend offered the following advice:
I'll tell thee what, Polly, that is all very well for those who have got a home and parents to shelter them, when work is slack; but depend upon it, many clever women find it, at times, a half starved life in those employments [mantua making]. Now, thee art a hearty, well-grown girl, and I think would be better off in service.Footnote 61
On the other hand, a cousin of Ashford's mother warned that entering domestic service would “injure my future prospects, as I could not be introduced into society by her or any of my respectable friends if I was a servant. […] I gleaned from her majestic oration that, in the event of my being out of place, they would have nothing to do with me”.Footnote 62 Overall, service provided young women with insulation from commercial depressions, making it advantageous for those lacking a strong support network. However, others saw service as disreputable, with some even threatening to withdraw support as a consequence.
Second, the opportunities presented by cottage industries and the textile sector were not substantial enough to affect servants’ wages. Cottage industries had been in decline since the mid-eighteenth century, limiting the ability of women to enter the sector.Footnote 63 Moreover, there is uncertainty regarding prevailing wage levels in these industries. While Robert Allen posits that high wages in hand spinning were pivotal to the early momentum of the Industrial Revolution in late eighteenth-century England, Jane Humphries and Ben Schneider provide estimates suggesting that wages in hand spinning were low and only grew slowly.Footnote 64 Importantly, shallow markets in cottage industries allowed putting-out agents to benefit from unequal power relations and culturally mediated wage-setting by employers.Footnote 65 Even so, there were limits to how many family members could be crowded into household-based cottage production. A recent assessment by Xuesheng You indicates that, even by 1851, the size of the female domestic service market exceeded that of all female manufacturing, and was four times larger than the number of women employed in woollen and cotton manufacturing.Footnote 66 Combined with the findings presented here, this suggests that these alternative sectors were not large enough to drive up wages.
The segmentation of the service labour market from other employment opportunities resulted in localised wage-setting, as indicated by the insignificance of the lagged dependent variable. This led to significant wage-setting power for employers. Allen notes that inequality was rising in the late eighteenth century,Footnote 67 and, indeed, in my analysis across columns 1–4 a one per cent increase in the log wealth of capitalists was associated with a decrease in servants’ wages of between 12.3 and 16.1 per cent. This association could have been driven by the reduced bargaining power of workers, distortions in the types of jobs created, and increased levels of market concentration. In environments characterized by high wealth inequality, concentrated economic power often allows employers to disproportionately influence wage-setting mechanisms, putting workers at a disadvantage. However, the relationship is not robust across all specifications when excluding observations from London, see columns 5–8.
In late eighteenth- and early nineteenth-century England and Wales, magnates played a major role in shaping local labour markets. Steve Hindle's research illustrates this, citing the example of Sir Richard Newdigate, a wealthy gentry landlord in seventeenth- and eighteenth-century Chilvers Coton, Warwickshire. At one point, Newdigate employed approximately six per cent of the local adult population as servants, in addition to those working in his estate's coal mines. Given staff turnover, a significant portion of the local population would have been either employed by or co-resident with Newdigate, demonstrating his considerable influence over the local labour market.Footnote 68 Here, it is worth briefly revisiting the findings of my study suggesting that proximity to London was associated with higher wages for servants. In London, the influence of a Newdigate-like figure would have been greatly diluted by the sheer number of wealthy individuals competing in the labour market in the capital.
Theoretically, more generous welfare expenditure could empower workers to negotiate for higher wages by raising their reservation wage. To test whether this hypothesis applied to the market for female servants, county-level log per capita poor relief expenditure was included in the models. As shown in Table 1, a one per cent increase in the log poor relief per capita was associated with a 15.9 to 16 per cent increase in servants’ wages. However, this was not robust when observations from London and Middlesex were excluded (columns 5–8). It is noteworthy that during the early nineteenth century, the primary aim of the Old Poor Law was to stabilize rural arable communities by providing out-relief payments to male agricultural labourers and their families, thereby counteracting land flight and ensuring a stable labour supply during periods of heightened demand.Footnote 69 The inconclusive results may reflect the dual nature of higher per capita poor relief, which simultaneously captures both levels of generosity and prevailing deprivation. For example, southern England was characterized by both high levels of relief expenditure and comparatively low female physical heights, indicating sustained pressures on biological living standards.Footnote 70
An examination of the time dummies reveals ongoing pressures on servants’ wages during the early nineteenth century. Table 1 shows that real wages decreased by approximately sixteen per cent between 1800 and 1819 compared to the end of the eighteenth century (columns 1–4). Wage pressures moderated somewhat in the 1820s, and wages returned to levels seen at the close of the previous century. Outside of London, real wages appear to have stagnated rather than declined (columns 5–8). Having established that wages were under persistent downward pressure during the early decades of the nineteenth century, the next section will explore how this relates to changes in wage bunching.
Trends in Employer Market Power and Its Sources
This section analyses changes in the bargaining power of servants and employers between 1780 and 1834, using wage bunching as a proxy indicator. It also identifies, drawing on qualitative sources, the factors that contributed to employer monopsony power, including localized labour markets, employer coordination in hiring, the absence of labour market protections, and servants’ reliance on character references.
The econometric results in Table 1 tentatively suggest the presence of employer monopsony power in the service labour market, coupled with sustained downward wage pressure in the early nineteenth century, which affected the bargaining dynamics between servants and employers. By analysing changes in wage bunching, we can gain insights into the prevalence of wage posting (i.e. when an employer sets a wage before meeting job applicants) and, by extension, employer market power. Of the wages recorded in the GLIH affidavits, 68.2 per cent were even numbers, while only 31.8 per cent were odd. Anecdotal evidence from household manuals indicates that even wage payments were instigated by employers. For instance, The Complete Servant, published in 1825, provides a list of recommended wages for thirteen different types of domestic servants. In 92.3 per cent of the cases, the suggested minimum wages were even numbers, while the maximum recommended pay was an even number in 84.6 per cent of cases.Footnote 71 In Mrs Beeton's Book of Household Management, a bestseller published in 1861, 60.2 per cent of the proposed minimum wages were even numbers, and 76.9 per cent of the maximum wages were even.Footnote 72 The predominance of even numbers in both household manuals and in the observed wage data suggests that these handbooks shaped wage-setting practices.Footnote 73
Bunched nominal wages represent a deviation from employer's rationality since wages are being set either too high or too low, thereby reducing profitability; this is a form of mis-optimization. In a competitive labour market, such mis-optimization should not occur because firms paying excessively high wages would go out of business, and those offering wages that are too low would struggle to attract workers. Recent research into historical and contemporary labour markets has linked the presence of number bunching to monopsonistic dynamics.Footnote 74 Findings by Arindrajit Dube, Alan Manning, and Suresh Naidu suggest that when wages are concentrated around round numbers for task-based work where productivity is easily measured, this may indicate wage posting. Similarly, Robert Hall and Alan Krueger argue that wage posting is prevalent in low-wage labour markets, particularly for unemployed workers.Footnote 75 Although the context here differs, the suggestive evidence of employer market power identified in Table 1, along with the anecdotal evidence of wage bunching in household manuals discussed above, suggests that the bunched wages of servants can be interpreted as take-it-or-leave-it offers by employers.
Moreover, since servants’ overall entitlements included both wages and in-kind payments, it can be conjectured that increased wage bunching coincided with a relative rise in the proportion of in-kind payments within the overall entitlement package. This idea is supported by Jordan Claridge, Vincent Delabastita, and Spike Gibbs, who find that an increase in the cash portion of workers’ entitlements during the Middle Ages was associated with greater bargaining power.Footnote 76 Thus, changes in wage bunching offer a useful indicator for measuring shifts in mis-optimization, and by extension, serve as a proxy for employer power in the service labour market.
To evaluate changes in labour market mis-optimization over time, Table 2 presents a logistic regression model to estimate the likelihood of wages being either even or odd. Again, eight specifications were run, and in all of them, the dependent variable was whether the wage paid was in an even number. The explanatory variables include the natural logarithm of the wage, the servant's literacy, whether the wage was paid in guineas (pounds being the reference category), the natural logarithm of the tenure of the employment relationship, the status prefix of the servant's employer, and the same set of four time dummies as Table 1. In columns 5–8, the time dummies were replaced with the log of the five-year average cost of living (CPI).Footnote 77
Table 2. Logistic regression of even wages.
Legend: County-level clustered standard errors in parentheses; *** p < 0.01; ** p < 0.05; * p < 0.1.
As shown in Table 2, higher wages were associated with an increased likelihood of even-numbered wages being paid, suggesting that wage posting was particularly prevalent in higher-paying service positions. A one per cent increase in the log of wages was associated with a 0.248 to 0.291 per cent rise in the likelihood of even wages being paid (columns 1–8). Furthermore, the time dummies revealed a clear pattern, with the payment of even wages becoming significantly more common between 1810 and 1834 compared to the period 1780–1799.Footnote 78 Servants in this later period were 8.8 to 12.3 percentage points more likely to be paid an even wage (columns 1–4).
By substituting the time dummies with the logarithm of the average cost of living (CPI) over the preceding five years, I estimated how changes in the value of in-kind benefits relate to the prevalence of employer mis-optimization. As Table 2 shows, a one per cent increase in the log of the average cost of living was associated with a 0.164 to 0.177 per cent rise in the likelihood of even wages being paid (columns 5–8). When the cost of the necessities of life increased, servants’ bargaining power over wages diminished.
As in-kind benefits were central to servants’ total economic entitlements, an increase in their relative value compared to cash wages raised the cost of unemployment. An unemployed servant would have fewer financial resources to offset the loss of in-kind benefits such as food, shelter, and warmth. This period of increased wage bunching coincided with significant declines in servants' real wages (Table 1) and in female physical heights, suggesting that young women who had experienced or witnessed severe nutritional stress within their families were possibly more risk-averse, prioritizing basic needs like food and shelter – the core components of a servant's entitlement package – over wages.Footnote 79
In addition to these economic pressures, legislative changes such as the 1824 Vagrancy Act created an increasingly repressive environment for those without a fixed abode.Footnote 80 As women's employment opportunities diminished with industrialization,Footnote 81 it appears that their bargaining power both within households and in the wider labour market weakened, enabling employers to misprice their labour more easily. The findings of Claridge, Delabastita, and Gibbs – that an increase in medieval workers’ bargaining power led to a higher cash portion and a lower in-kind portion of their compensation – appear to apply to female servants in the late eighteenth and early nineteenth centuries as well. When the value of in-kind payments rose, servants’ bargaining position weakened, reducing the costs for employers who mispriced their labour.Footnote 82
Having documented these changes in employer power, the autobiographies of servants and household manuals offer a useful source for examining the origins of that power and how it relates to the results presented in Tables 1 and 2. As the extracts above from the autobiographies of Rose Allen and Mary Ann Ashford show, relative housing and nutritional security were key attractions of service work. However, to access these benefits, women first had to secure a position. Both household handbooks and autobiographical accounts suggest that service labour markets were essentially localized, which granted individual employers considerable market power. For instance, Mrs Beeton's handbook advised prospective employers to recruit servants from “amongst her circle of friends and acquaintances, and her tradespeople” or through registry offices if other avenues proved fruitless.Footnote 83 Servants, on their part, relied on (female) peer and kin networks to find positions, with mothers playing a particularly important role in organizing placements.Footnote 84 When these peer networks did not yield opportunities, autobiographies indicate that the last resort of job searches would be enquiring at local shops.
Elizabeth Davis's account of arriving in early nineteenth-century London illustrates this. In her autobiography, she recalled that “having no alternative left, I then entered a shop in Tooley Street, and saw two men there, and asked if they knew of any family in want of a servant”.Footnote 85 She was subsequently referred to a tailor, who hired her as a cook.Footnote 86 These on-foot job searches support the findings of Ann Kussmaul and Charmian Mansell, who both argue that servant migration was predominantly over short distances. This aligns with the short-range mobility Colin Pooley and Jean Turnbull identify for the eighteenth and early nineteenth centuries.Footnote 87
By contrast, newspaper job advertisements, which required both literacy and the ability to attend interviews at the right time and place, provided a different avenue for employment. Elizabeth Davis described seeing an advertisement for a laundry maid in The Times and attending an interview:
There were sixty or more young women crowding two parlours on the same business. I was the tallest, and Mrs R--- asked me several questions over the heads of those who stood in front of me, and hired me a the rate of eighteen pounds a year.Footnote 88
With dozens vying for the position, it was the employer who set the terms of pay and conditions, notably offering Davis an even-numbered wage of eighteen pounds. Other autobiographical accounts suggest that wages were sometimes not discussed at all before employment began. Ashford recalled, “simpleton like, [I] never said a word about wages [when I started my employment]; and when the month [probation] was up, she [the employer] said I suited her very well indeed, and she should give me five pounds a-year, out of which I was to find my own tea and sugar. I made no objection then”.Footnote 89
The geographical limitations of kin networks and the distance women could walk in search of work restricted the number of potential employers, thus increasing their power. These examples contrast with how Jane Whittle describes servants’ wage formation in the early modern period, which she characterizes as “determined by personal bargains between two individuals, the servant and the employer”.Footnote 90 Rather than reflecting a bargaining process, servants’ wages often seem to have been determined by employer wage posting, especially in the higher-wage segments of the market and during periods of high cost-of-living pressures (Table 2).
To secure a position, a reference from a previous (respectable) employer was required. Mrs Beeton's handbook advised:
It is not well to be guided by a written one [reference] form some unknown quarter; but it is better to have an interview, if at all possible, with the former mistress. By this means you will be assisted in your decision of the suitableness of the servant for your place, from the appearance of the lady and the state of her house.Footnote 91
Additionally, the applicant was expected to arrange a meeting between the former and future employer:
The servant in search of the situation must be desired to see her former mistress […]. Your first questions should be relative to the honesty and general morality of her former servant; and if no objection is stated in that respect, her other qualifications are then to be ascertained. Inquiries should be very minute, so that you may avoid disappointment and trouble, by knowing the weak points of your domestic.Footnote 92
This approach highlights how employers could coordinate with each other on the conditions of servants despite the large number of employers in the market. Such collusion could result in wages and conditions being set at an industry level, without the need for a formal cartel. If employers followed Mrs Beeton's instructions (and the evidence of wage bunching suggests they did), a circle of colluding employers would form. One recent study of the late nineteenth-century Belgian coal mining industry found that similar employer collusion, particularly through cartel formation, suppressed miners’ wages.Footnote 93
A character reference was an indispensable asset for a servant. A denied reference could have dire consequences. Herbert Miller, an advocate for labour law reform in the nineteenth century, observed that “an employer has, in point of fact, the power to absolutely ruin even the best of servants, and the servant has no remedy [to a withheld reference]”.Footnote 94 As shown in Table 1, servants could not easily transition into other sectors. By withholding character references, employers could impose post-employment restrictions on servants, potentially locking them out of localized labour markets. Conceptually, this mirrors the effects of present-day non-compete agreements (NCAs), which prevents a person from competing with a former employer or business. Like NCAs, withheld references inhibited servants’ ability to move between jobs. In the United States, NCAs in low-skill sectors have been shown to suppress wages and reduce the impact of tenure on wage increases, as employers have less incentive to raise wages when workers cannot easily leave.Footnote 95
Miller further noted that “if she [the domestic servant] happens to be friendless, she has to stay and put up with anything her employers choose to impose upon her, because she knows that she cannot get another place without a character”.Footnote 96 Mary Ann Ashford's experience with her employer illustrates this: she endured “very bad living; very little meat, and the bread was kept till it was mouldy before it was cut”.Footnote 97 When she attempted to leave, “by giving warning, and going to other places; […] when the ladies came for my character, she [my employer] always said something that broke it off”.Footnote 98
There is also contemporary evidence linking workplace sexual harassment with the use of NCAs.Footnote 99 The sexual vulnerability of domestic servants is evident from court proceedings and their disproportionate use of institutions like the Foundling Hospital and Magdalen asylums.Footnote 100 In short, character references were a crucial source of employer power and created labour market friction by obstructing job-to-job transitions, a central factor in labour market outcomes.Footnote 101
Joblessness posed significant economic and social risks for servants. Mansell argues that, in the early modern period, young women outside of service were vulnerable to attacks on their character in church court hearings, reflecting the broader costs of housing instability.Footnote 102 While the costs of joblessness were high, legal protections for servants were increasingly eroded during the second half of the eighteenth century. In 1777, the High Court, headed by Lord Mansfield, ruled in in R. v Inhabitants of Brampton that domestic servants could be dismissed on account of pregnancy.Footnote 103 As Madeleine Chartrand shows, case law concerning the supposed “immorality” of workers disproportionately impacted female servants. Any sexual activity by a female servant, whether it occurred in her master's household or not, was grounds for dismissal, so long as she remained “in his family” (i.e. employed by him).Footnote 104 While the language of morality was ostensibly gender-neutral, pregnancy – the most visible sign of “immorality” – could result in discharge.
Beyond the stigma surrounding single parenthood, which Samantha Williams argues was prevalent, economic motivations also played a role in concealing pregnancy or childbirth.Footnote 105 Even when accusations of “immorality” were baseless, a servant needed legal knowledge and financial resources to pursue a wrongful dismissal case. Moreover, courts were not neutral; judicial proceedings were often influenced by the biases of class and gender.Footnote 106 In 1796, servants’ legal protections weakened further when the King's Bench ruled in R v Inhabitants of Hulcott that the rights stipulated in the Statute of Artificers applied only to workers whose occupations were listed in the legislation, a list that excluded domestic servants. As a result, domestic servants lost their legal right to recover wages.Footnote 107
Servants’ autobiographies indicate that non-payment of wages was a common issue. Mary Ann Ashford, in her first position at the age of twelve, faced this problem when her employer attempted to withhold her wages. After hearing rumours that her employer was involved in fraud, she decided to leave and later recalled:
When I left she did not pay me, and I called many times without getting it: as it was above a pound, Mr. Batt [a family patron] said I should not lose my first earnings, if he could help it; and he summoned my master to the Court of Requests, and I got it.Footnote 108
However, even if a servant successfully challenged a case of unpaid wages, the employer could still withhold a reference, thereby impeding future employment opportunities.Footnote 109 Not all servants has the benefit of patronage like Ashford, and the Courts of Requests only handled claims up to a value of £2.Footnote 110 Ashford also recalled that “my mistress deducted from my wages for every little thing I had broken during the seven months I had been with her”, illustrating how employers could make arbitrary deductions like “wear-and-tear” costs from wages.Footnote 111 The difficulty of recovering withheld wages was further compounded by tied housing, which formed a core part of a servant's compensation package, making it harder for servants to leave delinquent employers.
Rose Allen worked in a financially struggling gentry multi-servant household, where insufficient bedding during the winter, stale bread, and unpaid wages were recurring issues.Footnote 112 She recounted the crisis her colleague Alice faced who was wrongfully accused of theft. Alice was dismissed immediately, denied both her outstanding wages and a character reference. Without wages or a reference, Alice faced destitution and homelessness, only finding temporary refuge through Allen's offer of lodgings with her mother.Footnote 113 These dynamics reflect the processes behind the dramatic downward social mobility of unemployed servants, as observed by many of her contemporaries.Footnote 114
Servants’ autobiographies suggest that employers were well aware of their leverage over servants. Elizabeth Davis, for example, was employed as a general servant in early nineteenth-century Liverpool by a notoriously ill-tempered employer who frequently scolded and beat her.Footnote 115 After one physical altercation, Davis threatened to leave, but her employer taunted her, stating that she knew Davis had “no place to go to, as my cousin would not receive me”.Footnote 116 Other employers sought to control the interactions between departing and newly arriving servants to prevent discussions about wages and working conditions. Ashford, who left one position due to low wages, noted that her departure “sadly vexed” her employer. It was, she remarked, “one of those places that have got a bad name”.Footnote 117 Upon leaving, Ashford's employer insisted she depart immediately, as she made it “a rule that the one [servant] coming in should not see the one going away”.Footnote 118
The localized nature of service labour markets, coordination among employers, the importance of references, the composition of servants’ entitlement packages and the limited recourse to recovering unpaid wages, and the high costs associated with unemployment all combined to give employers significant power over wage-setting.
Conclusion
Labour has been identified as a key factor in driving industrialization. This paper contributes to the broader discourse on industrialization and its gender-specific consequences by analysing wage formation and the influence of employer market power in the late eighteenth and early nineteenth-century servant labour market in England and Wales. The findings indicate that limited and contingent money returns to tenure and literacy, alongside the lack of spatial correlation between wages and nominal wage bunching, point to the presence of employer market power. This power stemmed from localized and segmented labour markets, in-kind payments forming a core part of servants’ entitlement packages, employer collusion, inadequate employment protections, and dependence on character references. These insights enrich ongoing debates about the role of women during the early stages of industrialization.
While Robert Allen attributes Britain's industrialization to high wages,Footnote 119 the real wages of female servants – the largest occupational group – declined during the early nineteenth century, stagnating over the fifty years from 1780. From 1800 to 1820, female servants’ real wages and biological living standards deteriorated, with nominal wage bunching becoming more pronounced in the service labour market.Footnote 120 Women's bargaining power within both households and workplaces appears to have weakened simultaneously. Given the pivotal role of labour markets in shaping broader welfare outcomes, this paper's conclusion – that employer market power increased for the largest segment of the labour market during the early industrial period – calls for a reassessment of wage formation across the remainder of the eighteenth and nineteenth centuries, particularly its implications for gender-specific welfare.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0020859024000944
Open access
