Spanish subsistence wages and the Little Divergence in Europe, 1500–1800

Abstract

This paper suggests an alternative view of Europe’s Little Divergence in real wages. It presents a new dataset of prices and wages for Spain and proposes a new way of measuring the cost of bare-bones subsistence. The substitution of brown-bread prices for grain prices in the baskets transforms the scale and chronology of the divergence between North-western Europe and Spain. The results show that it began later and that unskilled subsistence wages in London and Amsterdam were significantly lower than those calculated by the canonical model, which would nuance the “high-wage” hypothesis.

1. Introduction

Connected to the Great Divergence debates, the so-called Little Divergence highlights the increasing economic divide between the North Sea area and the rest of Europe in the Early Modern period. Recent GDP-per-head estimates provide evidence of consistent growth in the Low Countries and Britain well before the Industrial Revolution, while in Spain, as in the rest of the continent, real incomes stagnated at best.¹ Real wages would reproduce similar patterns. Robert Allen’s computations of welfare ratios show that the fifteenth century was the last time when labourers enjoyed comparable wages all around Europe. After that, real wages diverged dramatically. While in North-western Europe they stabilised, or recovered after the decline of the sixteenth century, as was the case in London, unskilled workers on the rest of the continent saw their real wages plummet, albeit at different speeds (Allen 2001, 2009). Whereas the roots of the divergence have been traced back to late mediaeval times (Pamuk 2007; Allen 2011a), the gap would have increased significantly from the first half of the sixteenth century. At the turn of the nineteenth century, labourers in London and Amsterdam would have enjoyed real wages well above the minimum and several times higher than their continental counterparts. According to this literature, these high-wage economies were the outcome of the higher efficiency and dynamism of their economies, which had implications for trade, consumption, education, and health that would have paved the way to modern economic growth. In Allen’s words, “Britain’s unique wage and price structure was the pivot around which the industrial revolution swung” (Allen 2011b, p. 359).

However, recent research is casting doubts not on the existence of the Little Divergence itself but on this picture of an early and wide divergence. The main critiques to date discuss the quality and characteristics of the wage data used to compute welfare ratios. Malanima (2013) argues that London’s exceptionality would distort any comparison based on urban building wages and, in consequence, he chose southern England’s data in his England/Italy contrast. Geloso (2018) and Mocarelli (2019) suggest that the wage data for Strasbourg and Milan in Allen’s database underestimated workers’ actual earnings by not including other payments in kind. Even more far-reaching is the debate about wages in Early Modern England. Against Allen’s High-Wage Economy thesis, Stephenson (2018, 2019a, 2019b) maintains that wages in eighteenth-century London were substantially lower. Firstly, because in the series attributed to labourers, not all the rates corresponded to purely unskilled workers and, secondly, because these rates are extracted from the bills paid to contractors, which did not translate directly into wages paid to the workers. According to her calculations, daily wages were actually between 20 and 30 percent lower. Similarly, Humphries and Schneider (2019) provide new evidence in arguing that wages in spinning were generally lower and did not experience any substantial rise before the industrialisation began. Allen’s responses to these empirical challenges in defence of his high-wage hypothesis fuel the debate (Allen 2019a, 2019b).

While most of the dispute has focused on the cost of labour, we address the cost of subsistence. Allen’s theory combines low capital costs with high nominal wages and cheap sustenance. This would have given English workers higher consumption capacities than in other places (Allen 2009, p. 33). Nevertheless, we contend that the original subsistence line gives an imbalanced impression of the evolution of real wages in Early Modern Europe. Comparability becomes a critical issue in the debate on the extent and the timing of the European Little Divergence.

This paper makes two contributions. It follows up the data-revisionist historiography by presenting a new dataset of Spanish prices and wages that allows for more consistent guesses than those existing in the literature. Thus far, the leading Spanish reference in international contrasts has been Valencia, sometimes accompanied by discontinued calculations for Madrid.² However, Valencia is hardly representative of the whole of Spain, at least from the leading cities perspective. The main cities of the Kingdom of Castile would better represent the more dynamic urban centres of the country in the Early Modern times. In fact, the new evidence reveals the existence of two regional patterns regarding real wages: the Mediterranean pattern (Valencia), where in broad terms, real wages would have evolved according to the received wisdom about their evolution in Spain and most of continental Europe; and the Castilian pattern (Seville and Madrid). There, instead, the new data produce distinctive trends and higher real wages than those observed in the Mediterranean area and higher than those estimated for Madrid before (Allen 2001; Allen et al. 2012).

As a second contribution, the paper discusses an alternative way to calculate the cost of bare-bones subsistence in urban Europe. The canonical model computes the prices of the cheapest cereals available, as well as the less expensive legumes, meat types, and so on (Allen 2009, pp. 37–41). Instead, we propose a basket that uses the price of the cheapest type of bread available, defending the argument that it was not grains but bread that was the staple and the most important source of carbohydrates in urban Europe. Labourers in London, Madrid, Amsterdam, or Valencia demanded and consumed different sorts of bread and not cereals. The use of (brown) bread prices avoids the distortions that the inclusion of diverse types of grains may produce in the performance of subsistence wages. One of the arguments of this discussion is “the oatmeal effect”. We maintain that the use of oat prices artificially lowers the cost of minimum subsistence in London and Amsterdam, which explains a substantial part of these high ratios and the magnitude of divergence shown in figure 1.

Based on this new basket, we compare subsistence welfare ratios in Spain and North-western Europe. Our results show that, firstly, labourers’ real wages in the leading cities of North-western Europe were significantly lower than those estimated by the canonical model, and secondly, that the Spanish divergence would not be as early or as wide as previously thought. The brown-bread subsistence line proposes a different depiction of the Little Divergence and questions the high-wage argument from the perspective of the purchasing capacity of the daily wage, which may also have implications for some other narratives such as the Industrious Revolution (de Vries 2009a; Allen and Weisdorf 2011).

2. A new database of prices and wages for Early Modern Spain (1500–1800)

Until recently, the Spanish historiography was largely shaped by the long-term legacy of Earl Hamilton (1928, 1929, 1934, 1947) who bequeathed an impressive dataset of prices and wages. His New Castile series have been taken directly or indirectly as a representative for Madrid in the literature. However, Hamilton built them with data from different locations. Up until 1651, most of the prices came from the Hospital de Tavera in Toledo (Hamilton 1934; González-Agudo 2019). In the database for the years after 1651, again, prices from Toledo dominate, although many data come from a convent site in Casarrubios del Monte, midway between Toledo and Madrid. In total, only 10–15 percent of the quotations are from Madrid (López Losa 2013, p. 77). The price series of Madrid here are built with data from Hamilton (1934) for the sixteenth century and from Andrés Ucendo and Lanza García (2014) for the seventeenth century (1596–1700). The eighteenth-century price series comes from the data for Madrid in Hamilton unpublished worksheets and other primary sources.³ While the wealth of data on prices is impressive, the wage series lacked the same quality. The new series for Madrid stems from Andrés Ucendo and Lanza García (2014, 2015) for the years 1520–1700. The eighteenth-century data come from primary sources.⁴ This series provides more comprehensive evidence that compensates for the fragility of Hamilton’s series up to 1737. Compared with the dataset that Allen built based on the latter, this new data gives higher nominal wages for both labourers (on average 32 percent higher between 1601 and 1650, 21 percent for 1701–1750, and a 9 percent increase for 1751–1800) and masons (16 percent between 1651 and 1700, 18 percent in 1701–1750, and 19 percent in 1751–1800) (table 1).

The price series for Valencia comes almost exclusively from Hamilton (1934) and his unpublished worksheets kept at Duke University.⁵ Allen uses the same source in the spreadsheet of Valencia, but we have introduced some modifications in the dataset, the most important being the price of bread. We give new prices that differ substantially to Allen’s series derived from his bread equation. On the other hand, we use the equivalence of Valencia’s currency to grams of silver published by Feliu (1991) that show notable differences with Allen’s estimates, particularly in the eighteenth century.⁶ Both changes combined give a different impression of the evolution of the cost of living in the Spanish Mediterranean. Allen’s cost-of-living series for Valencia was consistently above the one for London for almost the entire period, and they would only converge at the end of the eighteenth century. Instead, our series has Valencia’s price levels falling behind those in London from the late seventeenth century. The new computation would better fit with the conventional wisdom of the chronology of Mediterranean decline and North-western ascent (table 2).

Evidence on wages in Valencia is again scarcer than on prices. Between 1628 and 1720, there are no data for labourers or bricklayers, except for 2 years (1661 and 1668) in Hamilton’s papers. There are some scattered wage rates for carpenters, painters, tailors, and confectioners. Allen used the last three wage rates to estimate the wage series of skilled workers and then those of unskilled labourers. Instead, we only took carpenters’ wage rates as a reference for skilled workers. Ledgers from the maintenance of the city walls provide complementary data that make it possible to fill some of the gaps⁷. The long-run stability of the skill premium in seventeenth-century Valencia (Lanza García 2014) served as a reference to extrapolate or interpolate the missing data. For the eighteenth century, wages come from Palop Ramos (1977a) and primary sources.⁸

This new dataset offers novel insights into the evolution of real wages in Early Modern Spain. Figure 2 compares, side by side, Reher and Ballesteros’s (1993) real wage index for New Castile with the new one for Madrid. Our index shows notable differences in levels and trends up to the late seventeenth century. First, it reveals that wage levels in central Spain were significantly lower at the beginning of the sixteenth century, and second, that the fall in wages was less pronounced than the old series implies. While the series by Reher and Ballesteros displays a long decline that lasted until the 1650s, the series for Madrid shows a clear recovery in the second half of the sixteenth century that extended into the first quarter of the seventeenth century, when real wages in Madrid would reach the highest levels of the whole period. From the 1650s, and amid the worst of the vellón inflation, the Reher and Ballesteros data show an unexpected rise to the 1680s. Instead, our series shows that real wages steadily fell from the 1620s to 1680. The differences that arise from the comparison of the series have a clear explanation: the quality of the wage series. Confronting the scarcity of wage data in Hamilton (1934) for the period 1501–1737, Reher and Ballesteros (1993, pp. 108–109) opted to elaborate the series adding wages and salaries from Valencia and Old Castile and later from Andalusia. Only from 1737 onwards could they use consistent data for Madrid.

Given the importance that real wages have in the demand-side approaches, we compare our new Spanish real wage index with the one that Álvarez-Nogal and Prados de la Escosura (2013) created for their long-run reconstruction of Spanish per capita GDP. Our comparison only covers the period 1500–1800 and thus provides a partial contrast. We give two Spanish estimates. Spain (a) replicates the model of the former. It computes the Spanish average with data from the kingdom of Aragón, represented by Barcelona and Valencia, and that of the kingdom of Castilla with data from Madrid and Valladolid from 1750 onwards.⁹ Considering the weight of Reher and Ballesteros’ series in the proposal of Álvarez-Nogal and Prados de la Escosura, the geometry of the series is quite similar to that described for figure 2. It shows significant variations in trends and levels up to the second half of the seventeenth century, being the years between the 1550s and 1630s the moment when the contrast is starker. Spain (b) includes in the computations for Castile data for Madrid, Valladolid, and Seville for the period 1501–1800. The inclusion of Seville transforms scales, particularly during most of the sixteenth century. The rise in wages after the court moved to Madrid in 1561 becomes diluted by the impact of the higher Sevillian remunerations and by the declining trend they exhibited for most of the century. By the last decades of the sixteenth century, the two new Spanish series tend to converge (Figure 3).

The last comparison deals with Allen’s computations of welfare ratios for Madrid and Valencia. In short, Allen’s methodology measures real-wage ratios concerning two hypothetical welfare lines: poverty (later, respectability) and bare-bones subsistence. In the canonical model, the cost of living for a bread-winner family of four is estimated at three times the value of a consumption basket that provides 1,941 kc plus a 5 percent allowance for the housing. The annual family income is computed by multiplying by 250 the daily wage of the male. We replicate this methodology to quantify the impact of the new Spanish price and wage data. Figure 4 compares the welfare ratios for these two cities as computed in Allen (2001, p. 428) and Allen et al. (2012, sec. Appendix 4, p. 31) with those derived from the new dataset. In this case, the comparison measures not only trends but also real wage levels regarding these theoretical welfare lines. As might be expected, the variations are not dramatic in terms of increasing (or decreasing) welfare ratios. Still, the new data give higher ratios, principally in the case of unskilled labourers. In Madrid, the most significant differences appear during the first half of the seventeenth century, with subsistence ratios 47 percent higher, and in the first half of the eighteenth century, when subsistence real wages would be 70 percent higher than in Allen’s data. In the case of Valencia, subsistence ratios are 60 percent higher in the second half of the seventeenth century, while the difference would be 35 percent across the eighteenth century as a whole.

Overall, the replication of the real wage indexes and the welfare ratios with the new data show developments and levels that sometimes contrast with the received wisdom, giving a less negative impression of the performance of urban real wages in Early Modern Spain, particularly in the sixteenth and seventeenth centuries. In the following sections, we discuss the contribution this Spanish dataset makes to the international comparative debates on the European Little Divergence.

3. Welfare ratios and the brown-bread basket: a new proposal for computing the cost of subsistence

The model of welfare ratios devised by Robert Allen constitutes an ingenious, simple but powerful tool for assessing the theoretical level of welfare that wages could achieve under some standardised conditions. This methodology has pushed comparative research beyond the contrast of trends, rates of change, or grain wages and has led to a fruitful global research programme. From its first version, the model evolved into two different welfare lines: respectability and subsistence. The first would correspond to average urban consumptions in Europe and was first labelled as the poverty line but later became a sort of model for respectable consumption.¹⁰ The second tries to calculate the cost of a theoretical threshold of subsistence measured in calories. It is based on the computation of the cheapest cereals available, with minimum allowances of animal proteins, and reducing other expenditures on non-food goods to the lowest possible level. Given its simplicity, the subsistence basket is becoming a global standard. As a model, it makes some bold assumptions to ease comparisons, which generate debate around its composition and results. The family of four, the number of working days, the budget structure and the bread-winner model are among the questions raised in the discussion. Allen (2001, p.472) himself recognised that the welfare ratio is a notional computation based on some arbitrary fixed assumptions that may change. Thus, it would be a mistake to understand them as an absolute measure of income or living standards. However, the respectability and subsistence lines provide benchmarks to evaluate specific levels of consumption, industriousness or economic development from a comparative perspective.

Some of the critical assessments have proved useful for improving the reliability of the model. Jane Humphries (2013) opened the discussion on the number of calories needed to sustain the ideal family. She argued that Allen’s figure of 1,941 kc per basket in the first poverty line, later adopted for the bare-bones subsistence line, would underestimate the hypothetical caloric requirements of a family when it is calculated as three times the needs of the male (1,941 kc × 3). Allen (2015, pp. 5–6) accepted this and came up with a new proposal, increasing the number of calories to 2,100 kc per basket and family consumption from three to four baskets (plus 5 percent each for housing), which increases from 5,823 to 8,400 kc the caloric budget. The impact of this change is significant, since the cost of subsistence would increase by approximately 40 percent. He asserts that London and Amsterdam would still be “high wage cities” because that change applies elsewhere. Therefore, the order of magnitude of the divergence between the North-western European and the rest of Europe remains unaltered. However, this modification would substantially reduce real wages and thus consumption capacities.

We join the debate on the methodology of welfare ratios by proposing a new way of computing the cost of minimum subsistence. On one hand, we adopt the new caloric criteria established after the Humphries–Allen debate. On the other hand, we propose using the price of the cheapest type of bread instead of grain prices. The idea is that a brown-bread subsistence line is better adapted to comparative research in Europe than one built with different types of grains. Firstly, it offers a more realistic assessment of the minimum subsistence. Secondly, brown bread is a homogeneous good that provides similar utilities elsewhere, something that is not the case when different types of grain enter the baskets. The combined effect of these two changes (the increase in calories per basket and the use of brown-bread prices) is a substantial increase in the theoretical cost of subsistence everywhere. However, as we discuss below, the rise would be significantly higher in North-western Europe than in the rest of the continent (table 3).

The case of “high-wage” economies rests upon a questionable premise. The original bare-bones subsistence line builds on the prices of the cheapest grain available. Hence, Allen used oat prices in North-western baskets, and rye, wheat, or maize were the alternatives elsewhere. We argue that the use of prices for oats brings the cost of subsistence down to well below what it would be if the price of wheat or rye in London and Amsterdam, respectively, were computed.¹¹ Moreover, it produces unbalanced comparisons and casts doubt on the real representativeness of the welfare ratios obtained for London and Amsterdam from a comparative perspective.

The measurement of a theoretical minimum subsistence can be approached in different ways. There is the linear programming alternative, which sets some nutritional standards in terms of calories, proteins, fats, and other nutrients and computes the cheapest way of fulfilling the established thresholds (Allen 2017). The application of the method is data-demanding since it implies choosing from a wide variety of foodstuffs to reach the established nutritional parameters. However, it is difficult to apply to its full potential in historical studies since data availability does not conveniently cover the requirements most of the time. This problem is particularly acute in the case of fruit or vegetables. Furthermore, because of their characteristics, the conventional sources available to produce price series (accountancy books of colleges, convents, hospitals, nobility houses, or even charity institutions) do not always fit with the reality of the consumption habits of the lower strata in the society. From qualitative sources, we know that the lower classes in Castile ate beef, not mutton, when they were consuming meat other than offal (Calderón Fernández et al. 2017, p. 348). In the Mediterranean area, mutton was dominant but cheaper kinds such as old sheep or goat meat and offal were widely consumed by the lower classes (Hernández Franco 1981; Caballero 1998). In the case of legumes, we find a similar case with chickpeas and broad beans, with the latter considerably cheaper than the former. However, the sources only rarely provide prices for such foodstuffs or for vegetables. It is undoubtedly also a problem elsewhere. Due to data constraints, from the four alternative computations of the least costly diets using linear programming that Allen (2017) describes, the bare-bones subsistence basket model would be close to the CPF diet. It would cover the minimum requirements of calories (2,100 kc), proteins (50 gr), and fats (34 gr) per day to maintain an active life. The subsistence welfare line is thus calculated with the cheapest foodstuffs from the set of provisions whose prices are available, providing similar utilities for all the places considered.

The biggest comparability issue we find in the computations appears in the type of grain. The choice here is crucial since it constitutes the main item in the subsistence budgets. The use of oat prices in the London and Amsterdam baskets would be in line with the basics of the model (the cheapest food of that kind). However, that premise is not fulfilled in the Madrid or Valencia baskets; here the choice is wheat, the most expensive cereal available in Spain. Rye, barley, spelt, buckwheat, millet, or even oats, together with maize in the northern Atlantic fringe or Valencia, would be cheaper alternatives.¹² The prices of the different types of grain oscillated over time but, in broad terms, the price relationship during the Early Modern centuries would be as follows: wheat, 1; rye, 0.75; barley, 0.5–0-6; maize, 0.5–0.7; and oats, 0.3–0.4. For its volume of production, barley could have been used to compute that hypothetical minimum of subsistence in cities such as Madrid or Seville, while maize could have been the alternative in Valencia. Substituting barley for wheat in the baskets for Madrid and Sevilla increases welfare ratios by 50 percent on average, while in Valencia, the use of maize prices would increase ratios by 46 percent. Another unsolved question is how the prices of these cereals (oats, barley) would have evolved if human consumption had competed with their use as animal fodder.

The alternative that we offer in this paper follows the idea of the cheapest way to reach the nutritional minimum but proposes the use of brown-bread prices instead of grain prices. In the absence of bread prices, the common choice in the literature is to use bread-grain prices in the composite price indexes. Additionally, bread-grain wages, measured as the litres that a daily wage could purchase, are the base for some of the first comparative approaches.¹³ Nevertheless, the use of grain prices involves other potential problems, as the political economy of grain trades may differ between countries, regions, and even cities. For example, van Zanden (2005, p. 176) found that increasing taxation on milling drove up rye-bread prices in Holland and he stated that “measuring real wages in terms of rye clearly gives a biased impression on their development”.¹⁴ Our working hypothesis is that porridges, soups, and boiled grains would have been part of the diets of many people all around Europe but that bread was still the staple for most of the urban labouring classes. Best-quality wheaten bread was expensive for rural and urban unskilled workers, but there were lower qualities of wheaten or rye bread that were more affordable. In England, oatmeal was widely consumed in northern regions, but less so in London or other parts of the south-east, where wheat was by far the most popular bread grain. Chales Smith estimated that by 1764, in the south-east of England, the consumption of bread corn was 89.3 percent wheat, 8.8 percent rye, and 1.9 percent barley, and it seems that this structure would not have been significantly different in the past (Ashley 1928, pp. 6, 25).The predominance of wheat over other bread grains began to be noticeable in London in about the fourteenth century, when cheap types of wheat bread such as “treat” brown bread began displacing rye or maslin breads among the poor (Campbell et al. 1993; Dyer 1998, pp. 199, 297–298). Wheaten bread, in its different types, became the staple for the labouring classes in London, as primary and secondary sources show. There, as well as elsewhere in south-east England, oats were primarily used as fodder for animals.¹⁵ Similarly, in Holland, rye bread was the staple for the working class in the Early Modern centuries (van Zanden 1993, p. 130; de Vries and van der Woude 1997, p. 626). Allen (2011a, p.11) argued that because their real wages were several times higher than the minimum that marked the oat-based basket, labourers in London or Amsterdam could upgrade their consumption to “white bread, beef, and beer”. One of the questions we will try to answer is how much room there would be to improve consumption after upgrading the subsistence basket to brown bread and how it would affect North-western wage levels in a comparative perspective. That is, by how much would real wage divergence in Europe increase, or decrease, when assessed in terms of types of bread of similar utilities?

4. Brown-bread prices

The supply of grains and bread at affordable prices was one of the main concerns of public authorities at every level in Early Modern Europe. In Spain, wheaten bread was the staple for the population in major cities and its price was a political issue of the first order. Thus, the intervention in the wheat and bread trades was particularly intense. Mechanisms were varied and affected selling prices in different ways. In Castile, the prices of grains had been regulated since mediaeval times by the Tasa. Although its real performance was widely debated, the Tasa lasted until 1765 when the grain trade was liberalised.¹⁶ Additionally, public granaries existed in many places, with the aim of securing supply and selling affordable wheat and flour to the local population, including bakers, at times of rising prices—which often entailed substantial financial losses. Finally, in many cities, the local authorities participated in fixing the maximum selling prices of bread, commonly with regard to the prices of wheat.¹⁷

It is difficult to quantify the real impact of these policies. In Spain, social upheavals were rare and this is commonly attributed to the public intervention to regulate wheat and bread prices (Castro 1987, p. 65). Feliu (1991, p. 33) found that shocks such as the Catalan Revolt of the mid-seventeenth century, the War of Spanish Succession in the early eighteenth century and the bread mutiny of 1789 tended to intensify the protection of the “poor’s bread”and caused its prices to increase to a lesser degree than those of white bread and wheat. The other side of the coin was the financial distress incurred as prices were subsidised with public funds much of the time. The authorities fixed bread prices in Barcelona by estimating that at least one-third of the bread sold in the city was white and its higher prices would compensate for any losses incurred from sales of brown bread. However, in times of scarcity, producers or the city council should cover the losses (Feliu 2013,2016). In 1767, the authorities in Madrid bolstered experiments to test whether the traditional way of setting prices corresponded with real production costs. The results demonstrated that in times of high wheat prices, bread prices fixed by the magistrates did not cover the cost of production of common and brown types. In this last case, the gap between revenues and costs could reach almost 30 percent (Campomanes 1768, pp. 224–225). Then, protection mechanisms would activate by financing prices of wheat, by allowing bakers to raise prices for the best sorts of white bread as a mechanism of compensation, or by directly subsidising reduction in prices for standard and brown bread. Llopis Agelán and García Montero (2011, p. 300) estimated that public subsidies to the Pósito in the last decades of the eighteenth century might have reduced the price of bread by approximately 8 percent but if they had concentrated in years when wheat prices were about 20 percent above the average, the price reduction would have reached a 25 percent.

Table 4 includes data on the purchasing capacities of daily wages in litres of grain and kilograms of bread and the percentage of the daily wage necessary to afford one kilogram of brown bread in each city. From a comparative perspective, the intuition of van Zanden would be correct and the averages of the years 1701–1725 present a good example: then, labourers in Amsterdam could buy 60 percent more grain (rye) and those in London approximately 20 percent more wheat than any of their counterparts in Madrid, yet they all bought the same quantity of coarse bread: approximately 10 kg. Purchasing capacities diverged in the second half of the eighteenth century but the situation did not change: in relative terms, the gap between the North-west and Spain was always higher in terms of grain wages than in brown-bread wages. Grain/bread ratios would describe the price difference between the raw material and the final product, and they appear to be lower in Madrid, Valencia, and Barcelona than in London and Amsterdam. In this last case, differences are influenced by higher production costs due to taxation on milling. In the case of London, one could argue the higher costs of labour as the reason for the higher wheat/bread ratios. Although evidence is scarce, it seems that the cost structure of producing wheaten bread was quite similar in Madrid and London.¹⁸ Our hypothesis points to the political economy of bread in Spain, the protective policies that Spanish authorities at all levels, from the king in Madrid to the town council in Barcelona applied to bread prices.¹⁹ It did not make bread in Madrid, Valencia, or Barcelona cheaper than in Amsterdam or London, but it reduced the impact of rising prices on population. Indeed, the bread purchasing capacity of the daily wage in Spain was lower, especially in the Mediterranean. In the case of Madrid, it was close to these in London or Amsterdam until the first quarter of the eighteenth century, but they diverged widely since then.

Brown-bread prices are computed as follows. In England, the Assize of Bread began setting prices in the Medieval times for several types of bread. In 1709, the Assize was reorganised and bread types reduced to three—White, Wheaten, and Household—according to their quality. They corresponded to the previous Wastell, Whole Wheat, and Treet types, and although their names changed, they maintained the same relationship concerning weights and prices, with Household being one-quarter cheaper than Wheaten bread. By 1758, the types of bread had been reduced to two: Wheaten and Household. They were of slightly better quality than in previous years, but the relationship still did not change (House of Commons 1774, pp. 47–48). In 1773, a third type was reintroduced, under the name of Standard Wheaten Bread. It resembled the Wheaten sort before 1758 but in this case with prices that were only one-eighth more expensive than the Household bread. If we measure quality in terms of flour extraction, the bread classes of the Assize were comparable with those observed in Spain (Table 5). Experiments carried out in the second half of the eighteenth century give extraction rates of 65–70 percent for white bread, approximately 75 percent for (standard) wheaten bread and 83–85 percent for household (brown) bread.²⁰ Prices for London derive from wheaten bread prices (Mitchell 1988; Allen 2001). We have converted them into household bread prices by reducing them by one-quarter. In Holland, through the Assize, city governments set prices for different wheaten bread qualities (from the finer white bread to brown bread) and rye bread, being the last one, together with the coarsest types of wheaten bread, the most demanded by the labouring class.²¹ The prices of rye bread in Amsterdam and unbolted wheaten bread in Utrecht come from de Vries (2019).²²

In Spain, wheaten bread was sold in various qualities and weights. The earliest evidence of the existence of different sorts of bread comes from the Mediterranean. In mediaeval Barcelona and Palma de Mallorca, sources describe three basic types: white, common, brown (Riera Melis 2016, pp. 102–106). In Castile, instead, the first available accounts are limited to generic statements about the baking of wheaten bread or the existence of lower quality types for the convenience for the urban poor. Only in the late sixteenth century did sources begin to provide more detailed information. In Madrid, although most of the quotations refer to high-quality white bread (pan de Corte) and common wheaten bread (pan de villa), there is news of brown types (pan de cabezuela) from the early seventeenth century at least.²³ In the eighteenth century, references to these three generic types became more frequent and the experiments of 1767 in Madrid offered the best description we have on the subject. The report describes the type and quantity of flour obtained per fanega of wheat, the pounds of bread baked, production costs, and market prices—as shown in table 5. Bakers in Madrid produced four types of bread: two types of the best-quality white bread, Candeal and French; one called Spanish, which was part of the group of common bread types; and the Town’s brown bread. Although less detailed, evidence of experiments in other Spanish cities gave equivalent results. In 1756, a test in Barcelona produced flour with a rate of extraction of approximately 60 percent for baking white bread; in the case of brown bread, it reached 83 percent (Feliu 2013, p. 13). In Valencia, in 1771 and 1793, the figures for brown bread were very similar: 85 and 86 percent, respectively.²⁴

These types of experiments are interesting for multiple reasons. On one hand, they reveal the characteristics of the types of bread on sale. On the other hand, they provide evidence about production costs and price differentials between bread types, which, along with the testimonies of the prices set by local authorities and other references from ledger books and literary sources, help produce reasonable guesses. Table 6 presents evidence gathered on price differentials between breads of different sorts in the cities of our sample.

The standard on which authorities established prices in Castile was the common bread, baked with flour at a rate of extraction close to 75 percent and giving approximately 85–90 Castilian pounds per fanega, depending on wheat quality. The customary practice was simple: the price in maravedís of two pounds of this type of bread equalled the price of one fanega of wheat in reales de vellón (rs vn).²⁵ Hence, if the price of the fanega of wheat were 20 reales, a two-pound common bread loaf would cost 20 maravedís. This rule provided a way to gauge prices of common bread in Castile. It also makes it possible to guess prices for coarser sorts of bread using similar methods. For Madrid, we followed two complementary methods that produced comparable results. When we had evidence on prices, we used the price ratios between the different types of bread, which were given in fractions of four maravedís called cuartos.²⁶ Differences between candeal and brown-bread prices oscillated between three and four cuartos of maravedí (12–16 maravedís) per loaf of two pounds. Thus, we have subtracted 3 cuartos to the recorded price of candeal bread and only used the figure of 4 cuartos when wheat prices were 20 percent over the decadal average. For those periods without evidence of bread prices, we estimated the price of brown bread from the price of wheat as explained above. The existing evidence shows that prices of brown bread moved in a range one-quarter to one-third below those of common bread. We have used the reference of one-quarter cheaper in our computations. In Valencia, the authorities carried out experiments routinely to ascertain the amount of flour and bread that six barchillas (100.5 litres) of regular wheat yielded and then set the price for brown bread accordingly. The abundant evidence available for the second half of the eighteenth century was used to build a new bread equation that substitutes the one proposed by Allen (2001, pp. 418–419).²⁷ In comparison, Allen’s bread equation produces higher bread prices, as much as 55 percent higher for the seventeenth century or 77 percent higher for the eighteenth century.²⁸ The series was tested using the bread/wheat ratios from Barcelona and other locations in the Mediterranean (Caro López 1985, 1987; Hernández Franco 1981; Feliu 1991), giving consistent results.

5. Spanish real wages in the North-western mirror

In his ground-breaking article, Allen (2001, p. 414) based his European comparison on what he defined as leading cities “to hold the side effect constant”. We have adopted the same perspective but adapted to Spanish circumstances. There are few doubts about the leading role of London and Amsterdam in their respective countries, but the question is more controversial in Castile. After Phillip II installed the Court in Madrid in 1561, the growth of the city was spectacular, yet it did not become the primary political, demographic, and economic centre for the kingdom until the beginning of the seventeenth century (Ringrose 1973, 1983). In the sixteenth century, other cities such as Seville, where, in the words of Braudel (1979, p. 65) “beats the heart of the world”, may have been more representative of the Castilian economic and demographic leadership. Thus, the Castilian leading-cities index is formed with data from Seville up until 1600 and then it uses Madrid’s ratios to 1800. In any case, real wages in both cities evolved in very similar ways and levels from the 1580s onwards. For the Spanish Mediterranean, we use the series for Valencia as it is the standard Spanish reference in the literature. Real wages in Barcelona moved along similar lines, and their inclusion would not alter the general picture.

Figure 5 displays subsistence real wages for unskilled building workers computed with our proposal of the brown-bread basket. It portrays an image that contrasts vividly with the scissor-like movement of wages in figure 1 in two respects: the chronology and magnitude of the Spanish divergence and the level of the ratios over minimum subsistence in North-western Europe. Subsistence ratios in Valencia do not deviate much from the conventional wisdom of an early Mediterranean divergence. However, the data for Seville and Madrid do not fit with the prevailing Spanish view and suggest a new reading of the evolution of real wages in Early Modern Spain. The higher volatility of prices and the monetary instability of the seventeenth century resulted in more significant oscillations in real wages in Seville and Madrid than in London or Amsterdam. While this is a relevant aspect to consider in the analysis (Sharp and Weisdorf 2012), differences in general levels do not show any significant divergence until the eighteenth century, when a fast relative and absolute decline in subsistence real wages occurred. Figure 5 and table 7 show that during most of the Early Modern period, real wages fluctuated differently, but in contrast with what is suggested by the canonical model, the gap between the high-wage cities (namely London and Amsterdam) and the Spanish cities was not so large.

Although the reformulation of the caloric content of the family budget raised the cost of subsistence, in Allen’s words, neither the degree nor the established chronology of the Little Divergence varies but the implied working year would increase by 40 percent and would offer less support “to the idea that an industrious revolution instrumented a consumer revolution” (Allen and Weisdorf 2011, p. 723). However, when subsistence ratios are estimated with brown-bread prices, the dimension of the divergence changes dramatically because, unlike the first modification, it affects the set of cities differently. All things being equal, the upgrade from oats to the cheapest type of bread increases the cost of subsistence by 88 percent in London and 87 percent in Amsterdam.²⁹ Instead, in Valencia and Madrid, the change to brown bread increases prices by only 22 percent. If the reference for comparison is the first set of estimates in figure 1, the amplitude of the variation is staggering, with brown-bread subsistence ratios now being 2.4 times lower in London and 2.6 times lower in Amsterdam. Real wages in these cities would still be the highest in Europe but the new figures vividly contrast with the distance from the minimum subsistence level that the canonical model estimated and that would have allowed labourers in the North Sea area to consume better and more varied goods and services than their European counterparts.

The brown-bread subsistence line transforms the scale and the chronology of the divergence. It was a more complex process than the original cut indicates, and the Castilian data provide a good example. Contrary to what the canonical model describes, our data suggests that labourers’ real wages in London and Amsterdam were lower and much closer to the bare-bones subsistence line. Also, they show that the differences in levels between North-western Europe and Spain were not so great and that the divergence began later.

6. Conclusion. Was the Little Divergence really so early and so significant as previously thought?

The Little Divergence is a well-established fact in the literature. The increasing divergence in incomes within Europe before the Industrial Revolution would correspond to the economic prosperity of the North Sea region in contrast with the stagnation seen in the rest of the continent. Research on the roots of North-western exceptionality has flourished in the past two decades but there is still discussion on the timing, sources, and magnitude of the economic divergence. In the case of real wages, Malanima (2013), Rota and Weisdorf (2016,2019), Malinowski (2016), Pfister (2017), and Stephenson (2019b) offer new and sometimes nuanced views of the divergence in real wages in Early Modern Europe. This paper aims to contribute to the debate with new Spanish data and with a proposal for computing the subsistence line using bread prices. It tries to answer two questions: what do the data presented in this paper offer in comparison with the prior estimates of Spanish real wages and to what degree is the received wisdom on the divergence in real wages questioned when we calculate the cost of the minimum subsistence in urban Europe with brown-bread prices instead of grain prices?

On one hand, the new data display a less negative impression of the performance of real wages in Spain. They show that subsistence ratios in Spain did not move very far from the subsistence line for most of the 1501–1800 period but that in urban Castile they were substantially higher than those computed before. These results add to the recent literature that points to a slightly better performance of the Spanish economy during the Early Modern period (Álvarez-Nogal et al. 2016; Llopis Agelán et al. 2018).³⁰ On the other hand, the issue of comparability is an essential aspect of the debate. We contend that the use of oat prices produces unbalanced comparisons. Here, we propose an alternative approach. We argue that it was bread that labourers demanded and consumed in the cities of our sample, and following the logic of the model, we build the cheapest bread basket available by using brown-bread prices. The brown-bread subsistence line transforms the scale and chronology of the European Little Divergence. According to our computations, during most of the Early Modern period, there were no significant differences in subsistence real wages between London, Amsterdam, and Madrid: only in the eighteenth century would the divergence become clear. Although we consider that the use of brown-bread prices is in line with the spirit of the “cheapest basket” model, we carry out two more simulations to test its reliability. The first is based on a basket computed with the prices of bread grains (195 kg of wheat/199 kg of rye). The second, with wheaten/common wheat-bread prices (271 kg). Figure 6 graphically summarises our proposal for the debate.

Figure 6.1 depicts what we define as the “oatmeal effect”. It compares London ratios, computed with an oat-based budget, and those of Seville/Madrid, calculated with wheat. Subsistence real wages in London would rise well above the minimum subsistence during the Early Modern centuries, and they would be significantly higher than the Spanish ones (as much as four times higher in the second half of the eighteenth century). However, doubts arise about the significance of the results because we contend that oat prices were less representative as a proxy for real-life consumption in London than wheat prices were for Valencia or Madrid. Figure 6.2 compares subsistence ratios computed with wheat (London; Seville/Madrid) and rye (Amsterdam) prices. In this case, the first visible change is the reduction in the scale of subsistence ratios in London. Figure 6.3 uses common wheaten bread prices, while figure 6.4 represents our proposal of subsistence brown-bread baskets. As expected, common and brown-bread subsistence wages were lower than those based on grain prices, bringing them closer to the subsistence line. However, the three graphs project a similar picture, which leads to two conclusions. Firstly, that the Spanish divergence only become significant from the early years of the eighteenth century. And secondly, that the substitution of bread, even the cheapest kinds, for oats in the budget dramatically reduces, in absolute and relative terms, the level of subsistence ratios in the leading cities of North-western Europe. Both conclusions nuance the magnitude of the Little Divergence and the high-wage economy hypothesis.

Supplementary Material

Supplementary material is available at European Review of Economic History online and at https://addi.ehu.es/handle/10810/40426?show=full.

Acknowledgements

Preliminary versions of this paper were presented at the Sixth Low Countries Conference, (Antwerp, 2011); the workshop “The global and long-term development of real wages: methods, problems and possibilities”, (IISH Amsterdam, 2012); the IV Encuentro AEHE (Iruñea-Pamplona, 2013); the Universidad Autónoma de Madrid (2014); the XV SEHA Congress (Lisbon, 2016); the ESTER-DATINI advanced seminar (Prato, 2016); and the 18th WEHC (Boston, 2018). The authors would like to thank Emiliano Fernández de Pinedo, Mario García-Zúñiga, Phillip Rössner, and the two anonymous reviewers for their comments on the earlier drafts of this paper. All remaining errors are our own.

Conflict of interest statement: None declared.

Funding

Financial support from Ministerio de Economía, Industria y Competitividad (Agencia Estatal de Investigación)/Fondo Europeo de Desarrollo Regional (FEDER), HAR 2016-78026-P, and the Basque Government (consolidated group IT897-16 UPV-EHU) is gratefully acknowledged.

Footnotes

References