John Hatcher, University of Cambridge Not many decades ago the long fifteenth century was a notoriously dark age in English history, neglected because it was located awkwardly between the ‘true’ middle ages and the early modern era. When at last it began to receive the attention it warranted, attempts to dispel the gloom were bedevilled by an ambition to fashion generalizations that fitted the whole experience of the hundred and fifty years after 1350, or even the quarter millennium from 1300 to 1550. As a result fundamental disagreements arose, the most notable being whether this era should be characterised by economic growth and prosperity or by recession and decline.2 However, contention cooled as more research was undertaken, topics on the agenda defined and prioritised, and more manageable chronologies adopted along with a willingness to identify sub-periods and sectors whose characteristics differed in major respects.3 Confidence has now increased sufficiently to persuade us that we are close to achieving a full understanding of the economy and society of England at the close of the middle ages, and there are distinct signs of a consensus emerging, with optimistic epithets such as ‘Economic Growth’, ‘An Age of Ambition’, ‘A Golden Age of Prosperity’, ‘An Age of Transition’, ‘A Consumer Economy’ and ‘A New Middle Ages’ in the ascendant. However, a little more probing reveals that there is much that remains mysterious about the era and paradoxical about attempts that have been made to describe and explain it. There is a library of economic, social and demographic theory that tells us what should have happened in the century and a half after the Black Death, but most of it fails to explain what actually happened. Many crucial elements of the pictures that have been drawn do not fit together as they should, and many leading indicators on which great reliance has are very unusual and in some cases contradictory. Attempts to incorporate the later middle ages into long-term models and data sets of economic and social change have also faced formidable difficulties. Much of the economic theory that has been applied to the era is ill-suited to the task because it was designed to analyse modern rather than pre-modern economies and the types of data that illuminate the workings and measure the performance of modern industrial and industrialising economies cannot be extracted with the same confidence from their pre-industrial equivalents.
Amidst all the uncertainties that remain, however, there is one crucial matter on which for centuries there has been universal agreement: the long fifteenth century was “a golden age” for the mass of the population. The promethean efforts of a host of compilers of wage and price data have combined to add depth and precision to observations first made by eighteenth-century political economists, and there is now a massive statistical archive that attests the prodigious levels attained by the wages of late-medieval workers.4 And these enthusiastic affirmations have been taken to confirm the existence of widespread levels of prosperity unparalleled for a preindustrial economy. Whether by association, intuition or economic theory, the fortunes of the unskilled have commonly been taken as a reflection of the exalted material living standards of the bulk of the population.5 Accepted wisdom has for centuries seen this era as characterised by living standards that were so phenomenally high they were not to be equalled until the mid-nineteenth century or even the 1880s, long after the benefits of advanced industrialisation had finally trickled down to the masses.
It is notable that the host of scholars who have diligently laboured in this field have arrived, without exception, at extremely optimistic outcomes. Thorold Rogers, the leading modern pioneer in the collection of price and wage data, was the first to use the term ‘golden age’, when he concluded in 1884 that ‘the fifteenth century and the first quarter of the sixteenth were the golden age of the English labourer, if we are to interpret the wages which he earned by the cost of the necessaries of life. At no time were wages, relatively speaking, so high, and at no time was food so cheap’.6 The publication seventy years later of the immensely influential Phelps Brown and Hopkins index duly revealed a doubling of the purchasing power of the day wages of urban building craftsmen between the first half of the fourteenth century and the third quarter of the fifteenth, and the most recent contribution, Gregory Clark’s computations based on a host of new wage and price data, shows the real wages of agricultural labourers soaring by more than 160% between 1300-49 and the peak reached between 1440 and 1479.7 Figure 1: Changes in the Equivalent of the Wage Rate of a Building Craftsman Expressed in a Composite Physical Unit of consumables in Southern England. 1284-1954
(from Henry Phelps Brown and Sheila Hopkins, A Perspective of Wages and Prices (1981), p. 19)
Figure 2: Real Agricultural day wages, 1209-1869
(from G Clark,’ The long march of history: farm wages, population, and economic growth, England 1209-1869’, Economic History Review, 2nd ser. lx, (2007), 109).
These are striking rates of increase, but even more extraordinary are the absolute levels that living standards are held to have attained in comparison with those of much later times. As Henry Hallam noted, with the approval of Thomas Malthus, in his View of the State of Europe during the Middle Ages, published in 1818:
‘There is one very unpleasing remark which everyone who attends to the subject will be induced to make, that the labouring classes, especially those engaged in agriculture, were better provided with the means of subsistence in the reigns of Edward III [recte IV] or of Henry VI than they are at present’.8
It was not entirely unexpected, therefore, that the peak of the Phelps Brown and Hopkins index reached in the later fifteenth century was not to be surpassed for more than four hundred years. There have been many adjustments and improvements to this seminal index, as well as a wide range of completely new statistical compilations that span the centuries from the middle ages to modern times, but the broad thrust of the original findings have been confirmed rather than supplanted. Indeed, the close similarities shared by the abundant series of long-run real wages that are now available lend great strength to their credibility. The experience of Clark’s agricultural labourers, for example, differs relatively little from Phelps Brown’s building workers when the purchasing power of their wages in the 1860s are shown to be still 20 per cent short of the exalted levels their fifteenth-century predecessors had achieved.9
Results no less stunning have been produced by reversing the standard statistical computations of what a day’s wage would buy by measuring improvements in welfare by the amounts of work required to provide subsistence. Farmer’s investigations of the money wages of threshers and reapers show that the number of units of work required to purchase a quarter of wheat and a quarter of barley fell by almost 60 per cent between the first half of the fourteenth century and the 1440s,10 and Allen and Weisdorf have recently computed that, at the money wages that are assumed to have prevailed in the fifteenth-century, an agricultural labourer in southern England would have been able to provide for all his basic needs, and those of a wife and two children, from no more than 200 days’ work.11 There is, therefore, a centuries-old and disciplines-wide consensus that the living standards of ordinary Englishmen and women at the close of the middle ages were stupendously high for a pre-industrial economy. This paper will challenge this consensus. It will argue that the towering peak in real wages that dominates the later middle ages in all long-term representations of living standards has been grossly exaggerated, and that it does not represent the course that the living standards of either the landless or the landholding majority followed.
Whereas there is no doubt that the real wages of labourers rose very substantially between the crowded and crisis-torn early fourteenth century and the spacious later fifteenth century, rates of improvement should not be confused with absolute levels of income. It will be shown that the real wages of the landless and near landless did not soar nearly as high as it has been conventional to believe, and that it is misguided to believe that the rate of improvement of the real incomes of the lower strata of rural society was representative of the experience of the majority, whose fortunes improved far less dramatically. Preliminary modelling indicates that the real incomes of the substantial body of middling peasants with holdings of roughly subsistence size remained relatively stable in the face of dramatic changes in the prices of labour and farm produce, while those of large-scale farmers declined. The golden age of the fifteenth century is in need of a severe dose of debasement.
Real wages are of considerable significance, not just for the fifteenth century but for all periods of history, and they are ensconced at the heart of almost all major descriptions and analyses of long-term economic, social and demographic development. However, it will be argued here that across much of history the conventional measures used to compute them are ill-suited to their task.12
The daily wage rates of workers and the prices of subsistence commodities are almost universally accepted as the data that can best perform the many crucial functions demanded of them. By the simple procedure of converting wage rates and the costs of subsistence into index numbers, and dividing the former into the latter, real wage indices are created that are precise and consistent, as well as abundantly available.13 They lend themselves admirably to the mapping of trends and fluctuations across centuries and continents and form the basis of comparisons of not merely the living standards but the economic performance of countries.14 However, despite their value in facilitating comparisons between economies and across time, there are many reasons why such crude data cannot bear the weight that is routinely placed on them and why their deficiencies proliferate and deepen the further back one goes in history. The most significant failings may be grouped under three main heads: first, the daily wage rates of labourers and craftsmen and the prices of subsistence goods do not constitute a sound basis for measuring the real incomes of these groups; second, the real incomes of landless or near landless males cannot be used as a surrogate for those of the landholding population at large, for whom daily wages were not the source of all income nor the market the source of all subsistence; third, the incomes of households are far more informative than the incomes of their male heads.
Henry Phelps Brown voiced these reservations. On the first page of his ‘Seven Centuries of the Prices of Consumables compared with Builders’ Wage-rates’ he stated in an unequivocal manner: ‘Nowadays, real wages are commonly estimated by comparing money wages with an index of the cost of living, but there are several reasons why we cannot do that here’.15 He elaborated, ‘all we have is the rate of pay for a day, we do not know how many days’ work the builder was getting in the year from time to time, nor what other resources he had.…and we know little or nothing about some important costs’. Phelps Brown continued, ‘These things apart, we still could not attach much meaning to “the cost of maintaining a constant standard of living” through seven centuries of social change. So we have not tried to construct any measure of real wages in the modern sense’. Accordingly, he took great care never to use the terms ‘real wages’ or ‘real income’ for his findings, and instead he composed the following lengthy title : ‘Changes in the equivalent of the wage rate of a building craftsman expressed in a composite physical unit of consumables in southern England’.
Yet, Phelps Brown’s admonitions have been repeatedly ignored by those who have used and interpreted his index, or amassed similar data to construct new indices.16 Time and again comparisons of the nominal daily wage rates of labourers and craftsmen with the prices of baskets of basic consumables have been deemed to constitute not just the real wages of those workers but their real earnings over a period of time, commonly a year. Indeed, their annual earnings are often computed by multiplying the daily wage by a working year of 250-300 days.17 This is not all, implicitly and explicitly the population at large is commonly assumed to have shared the same experience and benefited from the same rate of improvement in real incomes between the early fourteenth century and the later fifteenth. However, Malthus was not measuring the living standards of the average labourer, still less those of the mass of Englishmen and women, when he compared the money paid to a man for a day’s labouring and the price of grain in the fifteenth century with those prevailing in his own day.18 He was merely computing the quantity of food that a day’s labouring at that wage would purchase, and that is precisely what Gregory Clark’s newly-produced series of farm labourer’s wage rates and living costs does. The sources Clark is forced to use over the greater part of the period his series covers are the same as those available to all other price and wage historians, and they do not reveal how many days in each year the average labourer was employed at these rates, or what other sources of income and subsistence he might have had. There is also the vexed question of whether, when, and how much food and drink labourers received at work.
At first sight the real value of the pay received for a day’s work in the later fifteenth century might well appear comparable to that pertaining in the 1880s. But earnings are what really matter. A true comparison must take into account the number of days worked in the two periods, and there can be no doubt that this differed substantially. The labour market in the late middle ages, and indeed throughout the early modern centuries, offered far less regular employment than was to be the case subsequently when the great bulk of the population worked in non-agricultural occupations and the majority of those that remained in the countryside were employed on large farms.19 In pre-modern times paid employment in the countryside was characterised by its intermittent and piecemeal character: the changing routine of the seasons of the farming year combined with uncertain weather to cause wide fluctuations in the demand for labour and produce short-term and discontinuous working, and this inherent irregularity was accentuated by the fact that both the demand for labour and its supply came overwhelmingly from the occupiers of relatively modest farms. It is particularly regrettable that there is little or no useful information on the terms and conditions on which peasants hired each other because the likelihood remains that much work was undertaken between people in the same village using swap arrangements rather than cash payments.
In the fifteenth century access to land was unusually easy and the proportion of landless in the adult rural population was unusually low. The great bulk of the farmland of England was distributed in relatively small parcels among households that primarily used family labour to farm them, and if these households needed to hire additional labour, or had spare labour to market, they bought and sold it on an intermittent and short-term basis. The proliferation of large farms and the rise of an involuntary landless proletariat that came to dominate the agricultural landscape had yet to get under way. Relatively few villagers possessed lands that were sufficiently extensive to make it worthwhile hiring labour on a continuous and full-time basis, and those that did appreciated that their needs could be served better and more cheaply by servants hired by the year and remunerated with a combination of food, lodging, clothing and cash rather than by labourers hired by the day.
As far as the construction of a reliable wage series for the fifteenth century is concerned, it is particularly unfortunate that information has to be compiled exclusively from the demesne accounts of great institutions when by far the greater part of labour was hired by peasants and lesser lords,. Not only were these institutions unusually inflexible employers they progressively abandoned direct farming, so that by the 1440s the sample of accounts supplying data is tiny.20 There are also serious problems with the most dominant constituents of the agricultural wage series - payments for harvest work and for threshing and winnowing grain. These tasks were chosen because they are well-recorded and strictly defined, but they are also untypical and problematical. Harvest work was paid exceptionally well, commonly at double the winter wage, but such employment lasted for only a few weeks a year, while threshing and winnowing constituted only a tiny proportion of labour inputs in agriculture. Even on the largest seigneurial farms with expansive arable acres, threshing and winnowing the whole crop of grains and legumes rarely took more than the equivalent of the annual labour of a single man. Moreover, remuneration for this task was both lower and far more variable between manors than it is usually portrayed, and those that performed it were commonly paid by the piece rather than by the day.21
The daily wages paid for casual work during the slack periods of the year are of great significance but, unfortunately, this is when the great demesnes normally used their servants rather than hired labourers and the available data is consequently sparse, disparate and poorly-suited to inclusion in statistical tables. However, there is a body of robust detailed empirical evidence, much of it drawn from extremely informative farm accounts of lesser lords, that convincingly undermines the current wage series by demonstrating the prevalence of far lower rates of pay and the scarcity of employment for casual labourers during the dominant slack periods of the year. Poos’s detailed and resourceful analysis of the labour market in late fifteenth-century Essex, which he characterises as ‘highly episodic and discontinuous’, is a particularly enlightening case study.22 Unusually informative records reveal that the great bulk of the work running William Capell’s 300-acre mixed farm at Porter’s Hall, Stebbing, in 1483-4 was performed not by wage labourers but by eleven year-round servants, who were paid modest cash stipends ranging from 3s 3d to 22s 3d in addition to their board and lodging. Almost two-thirds of the days worked by the labourers hired at Porter’s Hall took place during the few weeks of harvest, when the males were richly rewarded with 4d per day and food. During the rest of the year, however, tasks such as sowing, weeding, ditching and harrowing required the hiring of far less labour and were remunerated at only 1-2d per day. Significantly, in the light of the heavy weight contributed by wages for threshing and winnowing in the series that estimates an average wage of 3.55d per day at this time,23 the four men who performed these tasks at Porter’s Hall earned an average of only 1.6d per day and laboured for only 51 days between them. Overall, the mean employment of the male labourers hired during the year on this farm amounted to just 7.8 days and their mean earnings to 24d. Of course, the records of a single large farm can give only a partial picture of the demand for labour in the region, and most of the labourers hired at Porter’s Hall must have found some additional employment elsewhere. However, Poos augments the intermittent employment and modest earnings of the labourers, cottagers and servants revealed by the Porter’s Hall accounts with complementary evidence from proceedings under the Statutes of Labourers and in wills and tax assessments.24 Such informative documentation is rare but not unique, and other sources attest to the prevalence of relatively low wages. On the Newton demesne in Cheshire at the turn of the fifteenth and sixteenth centuries, for example, casual labourers often received only 1d per day for unskilled work and harvest workers were paid merely 2d, and the same niggardly sum was paid to harvest workers at Millom in Cumberland. The Newton accounts, like those of Porter’s Hall, reveal a heavy reliance on servants employed on contracts, the cash element of which for males ranged from just 5s to 13s 4d.25 The archive of Durham Priory, which contains the names of employees, has enabled Christine Newman to complete a detailed study of workers and wages at the priory and its estates from 1494-1519, and she tells a similar story of brief, irregular and piecemeal hirings. Despite being by far the largest institution in the region the priory offered substantial employment to very few. Newman’s conclusion is that the golden age for labour was tarnished, for it did not have a ‘labour market characterised for most people by fixed employment, settled patterns or predictable career prospects’; most of the time workers were taking what they could get, which did not amount to much.26 This pessimistic view is shared by many other researchers, paradoxically including those who have compiled statistics that paint a far more optimistic picture.27
There are also special concerns about the authenticity of the summit of later medieval real wages which, according to almost all the series, was attained from c.1440 to c.1479. For this was a time when the country was in the throes of a deep and prolonged slump and the agrarian economy was enduring a most savage retrenchment. It is well-known that large-scale farmers faced severe problems from falling commodity prices and rising wages in later middle ages and that these led to the abandonment of demesne farming by landlords, frequent chronic indebtedness of entrepreneurs who ventured to lease abandoned demesnes, a shift towards pastoral husbandry and the relative scarcity and short-lived duration of large-scale peasant farms. In the mid-fifteenth century these adverse conditions profoundly worsened and engulfed substantial peasants and aspiring yeomen as well as the greater and lesser lords.28 It is not surprising, therefore, that the main driver that sent the already elevated real wage index even higher at this time was a further sharp fall in the prices of farm produce rather than a rise in money wages. More than this, there is evidence that the severity of the slump was reducing the demand for labour and driving employment and pay down.29
Thus, there is no substance in the belief that the average agricultural labourer in the fifteenth century was able to find employment for around 250 days a year at the exalted wage-rates recorded in various published series. Few labourers were able to find work whenever they sought it, or accepted it whenever it was offered, and it is unwise to assume that all of the work that was available was paid at the extremely favourable rates recorded in the published series. In the unlikely circumstance that a labourer was able to enjoy full employment at the 3.5-3.7d recorded in the Clark index, this lowly member of peasant society would have received an annual income of around £4, which is substantially higher than that of a senior full-time estate bailiff, who was paid a maximum of 60s annually, and only 20s or so below the declared taxable income of the majority of Warwickshire gentlemen in 1436 or the sum paid to the steward of the prior of Durham, who was a member of the gentry.30 But, even more significantly, such optimistic assumptions would have led to the perverse result that peasants could consistently earn far more from casual labouring than they did from working on their own land.
The calculation of the income tenant farmers received from working their own lands avoids many of the uncertainties and complications that dog attempts to estimate earnings from wages. If the calculation is restricted to arable alone it is able to rest on unusually robust data and non-controversial assumptions, for whereas real wage series cannot capture the participation rate or the earnings of either landless labourers entirely dependent on wages for their sustenance or the common run of peasant farmers who intermittently engaged in casual work, the physical output of an acre of arable farmland and its monetary value can be estimated with a considerable degree of accuracy, as can the number of days required to cultivate it. Of course, in practice there were no exclusively arable farms and farm labour was shared between growing crops and raising livestock, but the quality of the data and the robustness of the results make this theoretical exercise worthwhile, and for the purposes of illustration it will be based on the cultivation of crops on twenty arable acres, of which a generous fifteen acres are assumed to be under cultivation each year. Focussing on the period from 1450 to 1479, and using the distribution of crops given in Campbell’s medieval crop yields database, it is assumed that 5.7 acres of the twenty acres were under wheat, 4.8 acres under barley and 4.5 acres under oats, with the remaining 5 acres left fallow.31 Mean yields, net of seed, have been derived from the same database, and the selling prices of grains have been taken from Farmer’s series in the Agrarian History of England and Wales, 1348-1500.32 Together these data tell us that on average the gross value of an acre of wheat was 5.8s, after reserving seed for the following year, an acre of barley was worth 4.8s and an acre of oats just 1.8s. Thus, the total wheat crop would have been worth approximately 33s, the barley crop 23s, and the oats crop 8s, giving a combined sale value of 64s for the produce of the twenty acres. From this sum it is necessary to deduct rent and other seigneurial dues, estimated at 10s per annum, with a further 4s allowed for additional costs arising from the depreciation and maintenance of farm equipment, the costs of milling and suchlike. This gives a net average income of 50s per annum from the twenty-acres, before payment of tithes.
Meticulous calculations of labour inputs on the arable land of demesne farms by Karakacili and others indicate that from ten to just under fourteen days annually were spent on each acre, including the fallow.33 A relatively high figure of thirteen days has been adopted for these calculations, since peasants were likely to have invested more labour on their own holdings than did demesnes. Thus, our theoretical peasant farmer would have expended a total of 260 days working his twenty arable acres, which is conveniently close to the assumed full working year of an adult male, and each day worked would have brought him 2.3d.
This is a strikingly low figure, and if tithes are deducted from the crops the farmer’s earnings would fall to just over 2d per day, which is not much more than half the wage commonly attributed to casual labourers.34 Such a pronounced difference between the relative rewards of farming and casual labouring are very difficult to explain and would have been very difficult to sustain in practice. Apart from the fact that farmers would have been reluctant to pay labourers more than they added to the value of the output of the farm, any such gap ought to have been swiftly narrowed by a flow of labour from farming into labouring, which would have been relatively easy to achieve since the great bulk of the rural population commonly combined working for themselves with working for others.35 This is not all. Yet further doubt can be cast on the validity of common assumptions about wages and the availability of employment in the late middle ages by using the same methods and data to estimate the profitability of commercial arable farming. If all the conditions for the operation of the notional twenty-acres are held constant, excepting that now it is cultivated entirely by hired rather than family labour, the operation would have produced an average annual loss of around 30s between 1450 and 1479, before deductions for tithes. The prime reason for this dismal performance, of course, is the high imputed cost of labour. Even at a very low ten day’s labour per acre with no reduction in yields, the wage bill would have virtually matched the receipts from selling the complete crop, net of seed. While in practice temporary grazing on the arable would have been of some value to the farm livestock it would have fallen far short of that needed to meet the additional costs of rent, seigneurial dues, milling, capital depreciation of farm equipment, tithes and so on.