Economic inequality in the rural Southern Low Countries during the fifteenth century: sources, data and reflections

In recent years, the growth of interest in inequality has produced a wave of new empirical research on income and wealth disparities in pre-industrial Europe.1 The near general consensus emerging from this research has been that inequality tended to rise almost continuously and almost everywhere in Europe from the late middle ages until the 19th century.2 This remarkable result challenges earlier accounts of long-term patterns in inequality such as the Kuznets curve, and has found its way into recent overviews of historical inequality.3 A number of different explanatory factors have been suggested for this growth in inequality – with most recent literature emphasising the importance of regressive redistribution by emerging fiscal states, demographic growth, and proletarianization.4 Although those are valuable explanations for the Europe-wide tendency for inequality to grow, these broad causal factors do not provide a sufficiently detailed model from which to interpret local and regional differences in levels of inequality. Since contemporary and historical observers have traditionally discerned significant differences in social composition between different regions and localities, the apparent lack of a framework to


INTRODUCTION
In recent years, the growth of interest in inequality has produced a wave of new empirical research on income and wealth disparities in pre-industrial Europe. 1 The near general consensus emerging from this research has been that inequality tended to rise almost continuously and almost everywhere in Europe from the late middle ages until the 19th century. 2 This remarkable result challenges earlier accounts of long-term patterns in inequality such as the Kuznets curve, and has found its way into recent overviews of historical inequality. 3 A number of different explanatory factors have been suggested for this growth in inequality -with most recent literature emphasising the importance of regressive redistribution by emerging fiscal states, demographic growth, and proletarianization. 4 Although those are valuable explanations for the Europe-wide tendency for inequality to grow, these broad causal factors do not provide a sufficiently detailed model from which to interpret local and regional differences in levels of inequality. Since contemporary and historical observers have traditionally discerned significant differences in social composition between different regions and localities, the apparent lack of a framework to interpret these findings constitutes a lacuna in our understanding of pre-industrial inequality. Did the rise of the fiscal state, the transition from feudalism to capitalism, and demographic growth affect social disparities in cities and villages differently? And how did these processes affect regions and local communities to different degrees?
The current paper seeks to explore regional variation in levels of inequality in different types of localities and regions within the late medieval County of Flanders. We focus in particular on the fifteenth century, which was a turning point in the history of pre-industrial inequality, and continues to be the subject of debate on the prevailing levels of inequality in this period. In 1967 David Herlihy argued that in Pistoia (Tuscany), the Black Death had caused inequality levels to reach unprecedented heights. 5 Due to the partible inheritance system prevailing in the region, the severe epidemic fragmented landed property -initially giving rise to a more egalitarian distribution, but soon after resulting in an unprecedented concentration of property into a smaller number of hands. However, recent research on Northern Italy has now corrected Herlihy's view by showing that the egalitarian impact of the plague was more severe and longer lasting than Herlihy had imagined, and that as a result inequality was higher at the beginning of the fourteenth century (before the Plague) than at any time after that until at least the sixteenth century. 6 This has been attributed to the stronger bargaining power of labour after the demographic crisis of the fourteenth century, which is reflected in the simultaneous increase in real wages. 7 How long did the effects of the Black Death linger? Even though Herlihy's views on the effect of the Black Death on inequality have by now been confidently revised, there is much less consensus on how long the egalitarian effects of the epidemic lasted. Alfani's analysis of fiscal sources in North-western Italy has identified 1450 as a turning point, whereas later research on Tuscany indicated that there inequality started to increase again as early as the 1380s. 8 Remarkably, in both cases this growth in inequality predates the decline of real wages from the end of the fifteenth century onwards. Moreover, recent research has called into question the validity of the real wages data, suggesting that the 'golden age of labour' in the fifteenth century is little more than a statistical artefact 9 -thus casting doubt on the dominant explanation for the medium-term redistributive impact of the Black Death. Understanding when and where inequality began to increase in the century and a half after the Black Death, will allow us to gain better insight into the extent to which the Black Death caused a temporal reversal in trend, and into which factors ultimately undid these effects.
The County of Flanders presents a good case to study inequality in this key period. Although it was traditionally believed that Flanders was largely spared from the demographic catastrophe of the Black Death, this view has now been effectively discredited. 10 In the two centuries after 1348 the highly urbanized and prosperous County went through a period of economic reconversion: the traditional urban textile industries encountered increasing competition and the importance of international trade in the Flemish economy diminished. 11 Yet at the same time, this period has often been described also as a period of extraordinary 'Burgundian affluence' in the Low Countries. The fifteenth-century chronicler Philippe de Commynes praised the richness of the region as 'promised lands' ('terres de promissions'), the result of the benevolence of its ruler Philip the Good. 12 Quantitative data on the level of inequality in fifteenth-century Flanders is hard to come by. Earlier research on a consistent type of fiscal data from various cities in Flanders and Brabant, suggested that inequality began to grow from at least the sixteenth century onwards, but lacked sufficient comparable data for the fifteenth century to adequately pinpoint the start of this process. 13 Scattered research on fiscal data from Flemish cities has suggested that the late medieval economic reconversion caused an equalisation of income and wealth, resulting in more social stability, and a strengthening of social middling groups at the expense of the elites. 14 If this optimistic interpretation of Burgundian prosperity and social stability in late medieval Flanders is confirmed, it would provide an interesting contrast to the more pessimistic literature on late medieval Tuscany and Piedmont. However, we currently lack not only solid quantitative estimates of inequality to support this thesis, but also any information on the social patterns in the countryside. Since the relocation of industrial activities and changing patterns of access to land transformed the economic structure of the Flemish countryside in the late medieval period, it is to be expected that the late medieval crisis had different effects there compared to the city. Finally, if we are to better understand how specific social structures took shape in relation to economic and demographic changes, regional comparisons will be needed to identify potential causal factors. The current paper attempts to con-10 H. VAN  front these challenges by a) analysing inequality in a period for which we lack reliable quantitative data, b) concentrating on the countryside for which prior studies are largely absent, and c) setting up a comparative analysis.
DIRECT TAXATION: INCOME OR WEALTH?
The data presented in this paper is based on surviving tax lists from the fifteenth century. For rural settlements, these are usually the oldest lists available. For a number of communities older lists exist from the early fourteenth century, but because these only include a limited part of the population (those who owned land in the village where they resided), they have been excluded from our analysis. In this paper we focus specifically on the lists drafted for the collection of taxes raised by the state (i.e. the County of Flanders, ruled by the Dukes of Burgundy). Those taxes allow for a rough, but broadly reliable approximation of income inequality.
During the fifteenth century there was marked growth in both the frequency and level of state taxation. In the course of the fifteenth century the Burgundian rulers increasingly turned to their subjects to secure their income. 15 As has been shown for the counties of Flanders and Hainaut, state taxes were raised almost continually on an annual basis after 1450. The increased frequency of taxation was accompanied by a progressive rise in the level of taxation. In the course of the fifteenth century state revenue from taxes rose threefold in Flanders. The last quarter of the fifteenth century in particular stands out. During the 1470s and 1480s state extraction levels more than doubled compared to the third quarter of the fifteenth century. 16 The increased level of state taxation created a different fiscal dynamic in town and countryside. In the cities, the bulk of taxes was raised through indirect taxation. These taxes targeted basic and popular consumer items such as beer, grains and meat and therefore had a regressive character. In the countryside, on the other hand, there were no opportunities to raise taxes through indirect taxation. Rural communities had no other fiscal options than to resort to direct taxation.
In contrast to some other regions in Europe, wealth did not constitute the basis for direct taxation in the rural late medieval Southern Low Countries. There was no systematic registration or overview of the value of immovable and moveable assets of rural households. Only a few examples are known where wealth was specifically targeted as the object of taxation in the fifteenth century. These wealth tax initiatives were generally short-lived and exceptional. In the Duchy of Brabant, for example, a wealth tax was organised and raised in 1428. Households with assets worth less than 25 crowns were exempt. Those with assets worth between 25 and 4000 crowns paid a hundredth penny tax consisting of 1 percent of their wealth. with assets above 4000 crowns paid a flat rate of 40 crowns. 17 Because the tax did not produce the expected yield, the wealth tax was soon abandoned. Another rare example of a wealth tax comes from the county of Flanders. In 1440 the magistrate of the city of Courtrai obtained permission to organize a wealth tax to deal with the growing city deficit and debt. The numerous so-called outburgers ('buitenpoorters') of this city were also expected to contribute. They were rural dwellers who had obtained some legal privileges similar to the urban population. 18 In this particular case households were taxed at a relatively low level of 0.2 percent or 1/500 th of their wealth. 19 In both instances wealth assessments were based on sworn statements of the taxpayers themselves. Taxpayers stated their 'worth' before the local tax collectors and were assessed accordingly, but no systematic registration or valuation of assets accompanied these wealth taxes. 20 The exceptionality of state levies on wealth in the countryside suggests that taxation was more frequently targeted at income. As the extensive research of Antoine Zoete for fifteenth-century Flanders has shown, the so-called 'bedrijf' constituted the backbone of direct rural taxation in the late middle ages. 21 This can be roughly defined as the income generated by the economic activities of the taxpayers. In the countryside this more or less equated to taxing income from agricultural activities. As agricultural income was largely determined by the size of a holding, tax data broadly reflects differences in farm size. Although total landholding (and possibly the quality of the soil) was probably one of the most important factors determining individual tax assessments, taxes were not raised in strict proportion to holding size. When households enjoyed other forms of rural income these were also taken into account. For example, income from leases, tithes or annuities were also subject to taxation. Direct taxation also extended beyond the primary sector. Households with economic activities in the secondary and tertiary sector were also taxed in relation to their income. In the countryside this included innkeepers, millers, blacksmiths, textile workers, small traders and those active in the building trades. Although tax collectors aimed at assessing and taxing all rural income, in some cases the beneficiaries of this income escaped their grasp. Some groups enjoyed either full or partial exemption because of their social or legal status. More importantly, rural income transferred to urban dwellers through leases, tithes and annuities was not taxed.
Urban magistrates used their political power to erect a fiscal barrier between the city and the countryside. Direct taxation in rural settlements captured only the income that accrued to those who worked and lived in the countryside. Although income was undoubtedly the primary determinant of taxation, in some cases indications of wealth were used as a secondary factor to assess households. In theory, therefore, direct tax lists of rural settlements have some potential to analyse the distribution of income. 22

FROM INCOME TO TAX
Although most historians agree that income constituted the base for direct taxation in the late medieval countryside, tax lists cannot be treated as sources that reflect income differences and distributions perfectly. First of all, not all households that generated income were taxed. The largest groups enjoying tax exemptions were those deemed too poor to contribute. In the fifteenth century this was a substantial part of the population. In the county of Flanders a quarter of the households in rural areas and small towns were listed as poor in the hearth census of 1469. In the Duchy of Brabant some 30 per cent of the rural households were classified as poor in the fifteenth century. 23 In some cases, part of these households were probably supported by local welfare institutions. Most of the households labelled 'poor' in hearth censuses and fiscal documents, however, should be considered as those who were exempt from taxes. They did not necessarily receive support or alms, but their income was deemed too low to qualify for taxation because it probably just covered their basic subsistence needs. Therefore, only income above a certain level of subsistence was probably taxed. How low or high this minimum income threshold for taxation was, is difficult to establish. The large local differences observed in the relative share of the fiscal poor in rural communities and small cities suggests that no uniform practices were established and that village communities and their tax officials enjoyed some discretion in deciding who was included or not. Also, this threshold probably fluctuated depending on economic cycles and the tax burden. 24 The absence of any formal or uniform regulation on the exemption of the poor from taxes also suggests that communities enjoyed some discretion in this regard.
To complicate matters further, we have little or no information on how taxes were imposed relative to the income. Were incomes taxed proportionally or did tax collectors adopt regressive or progressive scales? At this stage we simply do not 22 In the course of the sixteenth century taxes were raised increasingly in direct proportion to the land used by taxpayers. See N know. 25 It is possible that a combination of scales was used in relation to the income level. What the extensive research on fiscal history for the fifteenth and sixteenth century has shown to date is that there were few internal conflicts in rural communities about the distribution of the fiscal burden. As the frequency and level of taxation increased, legal conflicts about taxation became more frequent. However, what was characteristic of these conflicts was that they rarely revolved around the relative contribution of taxpayers. Admittedly, such tensions and discussions possibly do not surface in the sources that have survived for this period. 26 This could suggest -but caution is required -that peasantries conceived the local distribution of the tax burden to be relatively fair and just.
One final point with reference to the reliability of tax lists as sources to approach income distributions, concerns the specific methods adopted by rural communities to capture the diversity in incomes. One striking feature of fifteenth century tax lists is that taxpayers are clustered in fiscal categories. In some cases this fiscal clustering seems exaggerated. In the tax list of the district Aardenburg (1475), for example, 435 fiscal units (i.e. households) are listed. These 435 fiscal units were divided over 20 different fiscal categories. 27 This suggests that variation in income in this community was relatively low as this variation could be captured by a mere 20 different fiscal categories. However, it is highly unlikely that this method accurately reflected actual variation in income. Within a given population eligible for taxation, we may assume that the variation in income was more diversified. Even within the same occupational groups we might expect some variation due to differences in wages, work intensity, household composition and developmental cycle, financial reserves and so on. In other words, the chances that households eligible for taxation will have enjoyed an identical income is close to zero. In order to test the extent to which late medieval fiscal data actually capture income diversity in a population we can calculate an index of fiscal dissociation. This index expresses the ratio between (1) the number of fiscal categories used by the tax collectorsand (2) the number of taxpayers/fiscal units. What we obtain is a value between 0 and 1, where 1 represents a situation of low fiscal clustering and high fiscal dissociation. In those cases the fiscal categories designed by the tax collectors have a strong tendency to capture income variation. On the other hand, if the index of fiscal dissociation reaches a value closer to zero, this indicates a greater degree of fiscal clustering: more taxpayers with presumably larger variation in actual incomes, were grouped in the same fiscal categories. In the example of Aardenburg cited above (where 435 fiscal units were grouped in 20 tax categories), the fiscal dissociation index is very low at 0.046. Graph 1 reports the outcome of these calculations for 24 fifteenth-century tax lists from rural settlements in Flanders. payers in these lists ranges from 4 in Koudekerke (K) to 435 in the district of Aardenburg.
The data suggests that in small communities the tax system could better reflect the diversity of incomes. There is only one settlement which reports a perfect match between the number of fiscal categories and the number of taxpayers. Unsurprisingly, this is also the smallest settlement: in Koudekerke four taxpayers are listed in 1417, 1420 and 1422. In all of these years there were also four fiscal categories. Although smaller communities were better equipped to capture these income variations, there were also differences between these communities. For example, Aardenburg (K) and Bredene (K) had a similar number of taxpayers during the first half of the fifteenth century. In Bredene (K) the number of taxplayers fluctuated between 22 and 24 between 1416 and 1420. In Aardenburg (K) 22 to 27 taxpayers were listed during the 1430's. Despite the similarities in the size of the tax base, differences can be observed in the number of fiscal categories. In Bredene taxpayers were divided into 7 fiscal categories, whereas in Aardenburg this number ranged from 14 to 18.

Graph 1. Index of fiscal dissociation
In general, larger settlements are characterized by lower fiscal dissociation ratio's. As the size of the taxed population supervised by a tax collector grew larger, it became increasingly difficult to tailor the fiscal system to the larger variation in incomes. This was probably the result of the workload experienced by tax collectors in large settlements to draft tax lists: the number of categories they used did not in-crease proportionally with the size of the settlement. In larger villages and parishes, drafting tax lists could be a labour intensive process and had to be repeated each year. During some years multiple tax list were drafted. Although local tax collectors were compensated for their work in compiling the lists, assessing the households and collecting the taxes, they may well have limited their efforts in establishing fiscal categories that fully captured the variation of income amongst the taxpayers in larger settlements. This pattern is confirmed by a comparison to a collection of 23 urban tax lists (from 8 different towns) from fifteenth-century Flanders. There the dissociation index was generally lower than in the countryside (averaging just 0,09), as larger populations did not lead to a proportional increase in the number of fiscal categories. In fact, in both town and countryside there seems to have been a clustering of most tax lists around the use of 20 to 25 different categories, except for the very small rural settlements. Working with more than 25 tax categories was probably too burdensome in practice, and might not have been considered necessary in order to arrive at a fair distribution of the tax burden.
Although smaller settlements were thus characterized by higher fiscal dissociation rates, there was also annual variation in how tax collectors constructed their fiscal categories. In Bredene (K) there were on average 3 to 4 taxpayers per category (see Table 1). The only exception seems to have been 1423. In that year the number of categories dropped to 4. Compared to earlier assessments this was a substantial reduction. This does not seems to have been influenced by the tax burden itself. Although total taxes paid in 1423 were lower than in the preceding year, the amount was identical to that of 1420 when 8 fiscal categories were used. Interestingly, a roughly similar picture emerges when we scrutinize the data on wealth referred to above. For the outburgers of the city of Courtrai we have information on their total wealth in 1440. This information was obtained by local tax collectors and assessments were based on sworn declarations from the peasants about their worth. The receipts of this wealth tax show that there was little variation in the declared wealth. 28 Overall, the 8330 outburgers reported a mere 78 dif-28 See Graph 2 at the end of this chapter. ferent wealth declarations. This implies that -on average -107 households within this group owned identical wealth. As can be seen from Graph 2, there was a clear clustering of wealth levels, especially in the lower categories. Out of 8330 declarations, an impressive 1604 or 19.25 per cent stated their worth at 25 lb. The three most frequent statements of wealth accounted for 48 % of all observations. It is highly unlikely that this accurately reflects actual differences in wealth. Rather, this form of 'wealth heaping' seems to have been influenced by a clustering towards numbers that could be used more easily in calculations. For instance, there was a clear clustering in numbers divisible by 12 or 20 -the subdivisions of the money of account in use. This indicates that rough statements of worth were more likely to be the aim, and probably much more pragmatically feasible, rather than very precise measurements of wealth based on actual quantitative data.
Low levels of fiscal dissociation (i.e. a low degree of differentiation between the incomes and wealth levels of households as the basis for tax distribution) were probably typical for the fifteenth century fiscal system and could also have been influenced by how contemporaries perceived social and economic structures. Also, the differences in the index of fiscal dissociation between settlements with a similar number of taxpaying units suggests that communities enjoyed some autonomy in drafting local tax lists. In the course of the sixteenth century land use came to dominate individual fiscal assessments. In the countryside most taxpayers would pay taxes in proportion to the land they farmed. As can be seen from the example of Westkapelle (K), this produced a more fine-grained fiscal system. During the years 1420, 1423 and 1439 the average (weighted) index of fiscal dissociation is 0.26. During the 1520's we witness a substantial rise of the index: from 1526 to 1528 the index rose to 0.45 on average. Relative to the number of taxpayers in this village, the fiscal categories in which they were assessed had almost doubled. This more fine-grained system of allocating fiscal burdens probably contributed to less opposition against these taxes, regardless of how income was taxed, and perhaps even to a sense of fiscal 'equity'. Also, this gradual change in tax assessments could hint at changing conceptions and realities of social structures in the countryside. For example, the higher Q3/Q1 ratio's for the sixteenth century indicates that fiscal middling groups were less homogeneous compared to the fifteenth century (see table 2).
One way to test the reliability of tax lists is to look for short-term variations. In the absence of large economic shocks we would not expect substantial annual variations in the distribution of income. For the village of Westkapelle (K) we have sufficient data to check for short-term changes in the distribution of income between 1420 and 1439 (see table 2). During the 1420's we see few short-term changes: inequality rose gradually, but at a slow pace. This relative stability of the inequality measures during these three years suggests that direct taxes on income were raised in a fairly consistent way. Between 1426 and 1439 the tax base declined by 14 units or circa 18 percent. This contraction was probably the result of increased mortality during these years as a result of outbreaks of the plague and harvest failure. 29 The volatility of inequality measures is much more pronounced during the sixteenth century compared to the fifteenth-century data. In contrast, the relative stability of inequality statistics in the 1420's suggests that income taxes were raised in a fairly consistent manner.

MEASURING AND EXPLAINING LOCAL INEQUALITY
Although Flemish fiscal records do not record and capture income perfectly, they can nevertheless serve as a rough measure of the distribution of income. 30 The absence of substantive short-term variations and the low overall volatility of inequality measures during the fifteenth century confirms the suggestion that these sources can capture the main structural features of the distribution of income. Although a lack of complimentary sources prevents a detailed check of completeness, there were few direct exemptions, and there is no reason to believe that a substantial part of the population was excluded from these levies. Although we must always tread with caution, it seems that the late medieval Flemish tax lists can be used to gain basic insight into economic inequality in rural settlements. Table 3 reports the results of inequality measures derived from tax lists of rural settlements in the county of Flanders during the fifteenth century, calculated from both published studies and unpublished sources. Although more tax lists have survived for his period, we have restricted the analysis to settlements with a minimum of twenty taxpayers. As can be seen from population size. Smaller settlements are represented by the rural enclaves of the ecclesiastical feudal estate (seigneurie) of the Chapter of Saint-Donatus in Bruges. In particular for the feudal sub-district Kanunnikse (indicated in the table by K) multiple lists have survived for different enclaves from the first half of the fifteenth century. 31 However, most lists contain too few observations to allow for the reconstruction of reliable inequality measures. Other tax lists contain a fairly large number of taxpayers and usually also include multiple parishes. In particular lists relating to a single administrative and fiscal sub-district (so-called ambacht) of the Liberty of Bruges usually contain taxpayers residing in two to five parishes. Although our research of the literature and intensive prospection of archival collections has resulted in a relatively abundant harvest of rural tax lists for the late 31 See Table 5 at the end of this chapter for a complete overview of surviving tax lists for this sub-district. medieval period, our data also suffers from a number of disadvantages. In addition to the small size of some of the settlements, the parishes for which data were obtained are all situated within a rather restricted geographical range. Most settlements included in table 3 can be situated within a 20 to 25 km radius from the city of Bruges. Only Kraaiwijk was located more to the west of the county of Flanders between Dunkirk and Calais (in present-day northern France). This geographical concentration also has implications for the diversity in social and economic structures under observation. Although the rural parishes around Bruges did not constitute a homogeneous group, there is little significant variation in their rural economic structures. Most parishes included in the study were beset by similar social, economic and demographic developments during the late middle ages. As extensive research during the past years has illustrated, this region experienced a profound restructuring during the late medieval and early modern period. Population declined dramatically in the long term, in part as the result of changes in ownership and distribution of land. The late medieval crisis, in this region exacerbated by ecological pressures, resulted in the transfer of ownership rights to land from the rural to the urban population. Because small peasant-owners could no longer support the heavy charges of dike maintenance, many were forced to sell their land to urban private and institutional landowners. These landowners increasingly leased their land to wealthy capitalist farmers. Therefore, this region displayed some of the basic characteristics of a rural society evolving towards an agrarian capitalist society.
In inland Flanders, on the other hand, we encounter a different social and economic structure. This part of the county was in large part still a region dominated by peasant holdings. 32 Unfortunately, these peasant communities are not represented in our observations. Our tax lists, therefore, do not reflect the variation in rural production systems and social agrosystems that characterized late medieval Flanders. A final disadvantage of this collection of tax lists is their distribution over time. For the fifteenth century there are no settlements where we can compare inequality measures at meaningful time intervals. Where such a comparison could be made (for example in Oostkerke) there are insufficient observations to guarantee the reliability of the inequality analysis. As a result, we cannot reconstruct longitudinal inequality measures for identical fiscal units during the fifteenth century. The dataset we constructed, therefore, is instructive to discern local and regional variation in inequality but does not allow us to reconstruct changes over time during the fifteenth century. 33 In line with previous research, we have restricted the analysis to a selection of four inequality measures: in addition to the widely used Gini coefficient, we have also included the share of the top 10 % and bottom 50 % of the fiscal distribution. To overcome some of the problems associated with the Gini coefficient we have calculated the ratio between the third quartile (Q3) and the first quartile (Q1). Be-  33 We can compare a number of settlements endowed with fifteenth-century lists with fiscal data from the 1530's and 1550's-1570's, but this analysis falls beyond the scope of this chapter. cause the Gini index is insensitive to changes in the income distribution in the lower middling groups, the Q3/Q1 ratio is a useful additional measure as it informs us about the income difference of the taxpaying population dispersed around 50 % of the median. In general, low Q3/Q1 ratios suggest that incomes of the fiscal middling groups were fairly homogenous On the contrary, high Q3/Q1 levels are indicative of large differences within this part of the fiscal population. 34 From table 3 we can observe that the Gini coefficient ranged between 0.42 and 0.52 in most communities. This is the case for large, medium and small settlements. The size of the settlement and taxpaying population seems to bear no relation to inequality levels. In general we find a strong correlation between the Gini coefficient and two other measures of inequality. In settlements characterised by high Gini coefficients, the share of the top 10 % is -as we would expect -more marked than in settlements with lower Gini coefficients. In most cases the Gini coefficient is negatively correlated to the share of the bottom 50 % of the distribution. However, there are also a number of settlements where the different inequality measures diverge. There are three settlements with Gini coefficients below 0.4: Bredene (K), Lombardzijde and Esen-Klerken. These settlements were also characterized by relatively low shares of the top 10 percent and relatively high shares of the bottom 50 percent. The low Q3/Q1 ratios are also indicative of a fairly equal distribution of income around the median. At the opposite end of the inequality spectrum we can identify Westkapelle (K), Aardenburg (K) and Kraaiwijk as settlements with high levels of inequality. In these three settlements Gini coefficients are higher than 0.55. This image of higher inequality is confirmed by the other measures. High Gini coefficients in these parishes are mirrored in the high share of the top 10 percent and low shares of the bottom 50 per cent. The latter was particularly marked in Kraaiwijk. Here, the poorest 50 per cent of the taxpayers only contributed some 5 percent. The two highest Q3/Q1 ratio's also pertain to two settlements with high Gini coefficients.
Although the geographical coverage of the data is limited, some notable differences between communities can be observed at the local level. If we look in more detail at the communities with low levels of inequality some patterns can be distinguished. Lombardzijde and Bredene (K) for example were both bordering the North Sea coast and situated close to regional centres of commercial fishing (Nieuwpoort and Ostend respectively). In the course of the fifteenth century, commercial fishing thrived in these settlements. For the village of Lombardzijde we know that during the fifteenth century a large share of households was directly or indirectly involved in the fishing industries. 35 For Bredene direct evidence is lacking, but given its proximity to Ostend, it is highly likely that at least part of the village community was dependent on commercial fishing for its livelihood. levels of inequality we encounter in these communities have also been observed in other regions. According to Kowaleski late medieval English rural fishing communities were characterized by egalitarian tendencies and a considerable degree of solidarity. 36 One of the factors that contributed to this equality was the specific strategy adopted by such communities to spread financial and operational risks. In fitting out fishing vessels, securing funding and attracting labour, small rural fishing communities resorted to a share system. In this system, crewmembers did not receive a wage, but were remunerated through a share in the profits of the voyage. The specific share of each crewmember on fishing vessels was determined by their investment in the capital, labour and equipment required to organize the fishing expedition. 37 In late medieval Flemish communities a similar system of shareholding existed. In late medieval sources those with a financial interest in the fishing enterprises were termed 'veynoten'. 38 Legislation also existed to prevent the formation of monopolies by restricting the number of crewmembers on individual fishing vessels and trips. This specific form of economic organization characterized by cooperation and anti-monopolistic regulation is likely to have produced more egalitarian outcomes. The low inequality levels observed in the fiscal data for the coastal fishing settlements in table 3 reflects this pattern. This picture of low inequality is supported by archaeological evidence for this region. There were no substantial differences, for example, in the size and quality of houses in these rural fishing communities. 39 Whereas fishing might explain low levels of inequality in Lombardzijde and Bredene (K), we must turn to other explanations for Esen en Klerken. These settlements were situated further inland and had no connections to maritime enterprises. They were located in the western part of the Liberty of Bruges and south of the small city of Diksmuide. Esen and Klerken both bordered the forest of Houthulst, the largest serried woodland area in late medieval Flanders. At the start of the fifteenth century, the forest encompassed a total surface area of more than 50 km² and bordered seven parishes (including Esen and Klerken). By the late middle ages the forest was co-owned by the count of Flanders and the abbey of Corbie who each had an equal claim on ownership and share of the profits. Income de-rived primarily from the sale of timber. 40 Although the woodlands were privately owned, the inhabitants of the nearby seven parishes (called 'ommesaeten') enjoyed considerable use rights in these forest as recorded in the custumal compiled during the first quarter of the fifteenth century. In exchange for a low and fixed annual payment inhabitants of these parishes could claim access to the forest and its resources. Villagers could pasture their cows in the forest from May to the end of the harvest period. It also provided an important source of energy for the villagers, as they could cut peat and collect firewood. Finally, the communities bordering this woodland complex also enjoyed the right to cut sods. These sods were mixed with animal excrements and served as manure for their arable land. 41 The extensive use rights of the villagers enabled them to derive income from land beyond the limits of their own holdings. Also, use rights to the forest were not socially selective; all inhabitants of the seven parishes enjoyed equal access to its resources. The equality villagers enjoyed in terms of their use rights to the forest seems to have been reflected in a low level of general economic inequality. The very low Q3/Q1 ratio indicates that income differences within the fiscal middling groups were small. Although there was an economic elite in these villages, the economic distance between this group and the rest of the population was limited. This seems to suggest that common use rights in these settlements exerted a downward pressure on economic inequality. Nevertheless, it should be stressed that at this stage it remains difficult to identify the mechanisms that produced this relatively equal society. Inclusive common use rights as such do not always produce more equal societies, as in some cases they could also reproduce existing inequalities. It is far more likely that the roots of low income equality in Esen and Klerken stretch further back in time. Possibly, when parts of the forest were cleared in previous centuries and transformed into private land, the owners adopted a policy of fairly equal distribution. Such a pattern has also been attested in medieval northern France. 42 In this case, the extensive use rights simply maintained and reproduced the original equality in access to land. Moreover, the inclusive use rights erected an institutional barrier to the concentration of economic resources. Elites were constrained in their ability to claim a larger share of rural income because use rights were inclusive and protected by custom. Also, these use rights may have made villagers more resilient to economic shocks and cycles. In any case, both examples of villages with low levels of economic inequality suggest that there was a clear link with the nature of social and economic organization, even though the exact causal mechanism remains difficult to uncover for now. At the other end of the spectrum we encounter a number of settlements with relatively high levels of inequality. As we can see from the data for Westkapelle in the fifteenth and early sixteenth centuries, this relatively high inequality was a structural feature of some of these villages (see Table 1). In contrast to the villages with low inequality discussed earlier, there are no indications that settlements like Westkapelle or Aardenburg were endowed with specific social and economic structures that would result in higher inequality. As noted above, overall high levels of inequality are consistent with what we would expect from the results of historical research on this region. From the fourteenth century, this region was transforming from a peasant society into a region dominated by large leasehold farms. Local differences in inequality could, therefore, signal that we are observing settlements during different stages in the transition process. In villages with higher inequality the forces of agrarian capitalism had already profoundly permeated and shaped social and economic structures. Differences in equality measures for these villages, therefore, could reflect differential developments in speed and intensity of the transition process. As the data in table 3 illustrate, such differences at the regional level could be substantial. In Kraaiwijk economic inequality was pronounced: all the inequality measures point to an extremely polarized income distribution. In other villages of coastal Flanders, inequality was markedly lower. Although recent research has convincingly argued that villages in this region were characterized by a number of shared characteristics, inequality statistics indicate that the speed and intensity of this transition was far from equal. 43 Clearly, there were a number of factors accelerating or slowing down the process of agrarian capitalism. As agrarian capitalism in this region was largely driven by the concentration of landownership in urban hands, and the ecological pressures caused by flooding, both the proximity of cities and of flooding-prone water could influence this. However, in the absence of data on landownership and flooding intensity and frequency at the local level their impact is difficult to evaluate. Also, we should consider the possibility that our sources simply do not capture and reflect income differences in these villages. In a highly proletarianised society, those at the lower end of the income distribution would not have contributed to taxation and therefore, will not appear in the tax rolls. Possibly, the presence or absence of labourers in these lists also influences the variations in income inequality observed in this region. Finally, we should also consider the possibility that agrarian capitalism resulted in more egalitarian societies at the local level. If, as a result of farm engrossment, labourers were forced to emigrate, they would disappear from our observations. Although they would remain essential for large farmers, they could well have lived outside the communities we are observing. There are some indications that fifteenth-century farmers relied on migrant labour recruited from regions further inland. If labour was simply not present in these settlements, the outcome would have been a more equal society. 44 One of the main conclusions that we can draw from our comparison of villages within Flanders is that the region was characterized by different inequality regimes. We observe a substantial variation in the inequality levels between the settlements. In-depth research on some of these village economies can certainly offer some valuable clues to explain these differences. However, in the absence of sources and data on the wider social and economic characteristics it is not always possible to identify the forces that shaped local inequality levels. Our overview of inequality in the Flemish coastal area illustrates that villages can experience substantial differences in their inequality levels under identical ecological, economic, social and demographic conditions. The data for coastal Flanders indicates that the transition to capitalism in the fifteenth century did not generate identical and interchangeable village societies and social structures.

INEQUALITY IN CONTEXT
The relative richness of late medieval fiscal data for the county of Flanders is demonstrated by the scarcity of sources for other parts of the Southern Low Countries. For other regions, only a small number of tax lists from the fifteenth century have survived, in particular for rural settlements. 45 Options to compare inequality statistics at the regional level are therefore limited. However, because the few existing tax lists from other regions come from villages with different social and economic structures compared to those in coastal Flanders, they are crucial in contextualizing our results, and in order to provide a limited degree of comparison. For the village of Rijkevorsel in the Duchy of Brabant we dispose of six tax lists from 1464 to 1475. As in Flanders, these lists were drafted to collect state taxes. Rijkevorsel is situated to the north-east of Antwerp and was part of the Campine region. Communities in this region were characterized by extensive heathlands held in common property by the rural communities. 46 The share of common land within these communities was usually substantial: in many villages 60 to 80 per cent of the surface consisted of common land. 47 Importantly, the commons in the Campine have been characterized as 'inclusive'. The vast majority of the population in these villages enjoyed access to the common lands and its resources. Unlike in other regions, common resources were shared relatively equally among the population and therefore contributed less inequality, or even helped to diminish it. A comparison of the inequality results for the village of Rijkevorsel with those of coastal Flanders nevertheless reveals a number of surprising similarities (see table 4). In the village of Rijkevorsel tax lists from the second half of the fifteenth century report Gini coefficients within a relative narrow range of 0.59 to 0.63. As such, they are similar to the highest levels of inequality recorded in the villages of coastal Flanders. Also, the share of the top 10 percent is in most cases higher than in coastal Flanders. Finally, Q3/Q1 ratios are also indicative of a more polarized fiscal middling group. The inequality measures calculated from these fiscal sources suggest that inclusive commons did not necessarily produce more equal societies than those characterized by agrarian capitalism. On the contrary, inequality was marginally higher compared to villages with no common land, privatized use rights and an ongoing process of farm engrossment. Also, there was a marked difference between Rijkevorsel and Esen-Klerken. The inhabitants of these latter villages enjoyed forest use rights and were characterized by low levels of inequality. Although the inhabitants in both Rijkevorsel and Esen-Klerken enjoyed access to either common land or common use rights, this was no reflected in similar inequality statistics.

Tab. 4. Inequality statistics for Rijkevorsel and Hoves-Graty, 1464-1475
Year A second community that can serve as a point of comparison is the village of Hoves-Graty in Hainaut. Situated in the north-eastern part of the county, Hoves-Graty was located in one the most advanced and densely populated regions of Hainaut. 48 In terms of social and economic structure, Hoves resembled a more traditional peasant society. In this region farms were relatively small (3 to 5 ha) and aimed at self-sufficiency. There were a couple of larger leasehold holdings where wage labour was employed, but their number was limited. One of the characteristics of this village and the wider region was the importance of rural textile production. Within these villages we encounter both specialized textile artisans, but also proto-industrial production in peasant households. 49 In terms of its social and economic structures, Hoves-Graty resembled the part of inland Flanders bordering this region to the north. For Hoves-Graty we have two tax lists dating from the fifteenth century (1465-1466 and 1470) from which we can calculate inequality statistics. 50 Importantly, the inequality measures demonstrate some volatility during the two years when we can reconstruct income inequality. Between 1465 and 1470 the Gini coefficient grew by some 20 percent. This suggests that the fiscal system in Hoves was probably less stable compared to the other villages for which we have data. Hoves is the only settlement where we encounter such substantial variations in inequality measures during the fifteenth century. Nevertheless, the data are still instructive from a regional perspective. Overall, inequality was lower than in Rijkevorsel. All inequality statistics point to a more polarized distribution of income in the Campine area. A more typical peasant society seems to have been less unequal than a community with inclusive commons. Such a peasant village like Hoves was also highly similar to some of the villages of the coastal plains. Based on their inequality statistics, Hoves in Hainaut and the district Aardenburg in Flanders were characterized by a fairly similar distribution of incomes.
Admittedly, the number of communities where we can systematically compare economic inequality and their underlying social and economic structures is limited. Therefore, caution is advised in drawing strong conclusions from this comparison. However, similarities in terms of inequality between different regions can also be observed during later periods. Fiscal data for the sixteenth century indicates that villages in Flanders and Brabant characterized by either agrarian capitalism, peasant agriculture or inclusive commons produced comparable levels of inequality. 51 Restricting the analysis to inequality measures in the 1460s, we would be tempted to characterize Rijkevorsel, Westkapelle and Hoves as rural settlements with a number of shared social and economic characteristics. There is a marked dissociation between the social and economic structures characterizing these villages and the inequality levels we observe in the fifteenth century. The traditional parameters used in rural history to characterize and analyse village communities in depth (farm size, landownership, common land, use rights, social structure, wage labour) do not predict the outcome in terms of economic inequality we would be expecting. A regional analysis and comparison of rural economic inequality in the Southern Low Countries, therefore, results in a rather unexpected outcome.

CONCLUSION
In the absence of longitudinal data and sources, this paper focused primarily on local and regional differences in economic inequality during the late medieval period. A number of tentative conclusions can be formulated based on our anaysis, which can at least inform future research on this topic. First, at the local level we can observe substantial differences in economic inequality within a restricted geographical range. We have argued that such local differences in rural income inequality resulted from local differences in agrarian systems of production. As access to land and other agrarian means of production could be subject to highly varied local conditions, we can expect larger variations in inequality in a rural context. Although economic inequality in urban settlements was higher, differences in income distribution between cities were perhaps less pronounced than between rural settlements, during the late middle ages and early modern period in the Southern Low Countries. 52 In contrast to urban settlements, the aggregation of inequality data for rural settlements to construct regional data poses more challenges. As our analysis shows, even settlements with a shared set of social and economic characteristics sometimes display different levels of inequality.
Second, our analysis has also shown that the regional comparison of income inequality exposes a number of unexpected challenges. Inequality measures do not reflect the regional differences in rural production structures which previous research has shown to be in place in late medieval Flanders, Brabant and Hainaut. It appears counterintuitive to find similar levels of economic inequality in villages experiencing a rapid transition to agrarian capitalism, peasant communities and inclusive commons. A regional comparison of inequality measures mainly brings differences in degree to the fore, whereas traditional analyses of rural economic structures have often exposed and stressed fundamental differences in kind. Inequality statistics derived from fifteenth-century fiscal sources, it would seem, tend to downplay the importance of regional specific rural production structures in determining the distribution of outcomes. A number of factors can explain this discrepancy. First, the nature of taxation could have been different between the regions. As we argued above, the absence of any strict or detailed regulation on taxing rural household income means that communities had some autonomy in deciding who and how to tax. This means that rural communities could have adopted a different minimum threshold to tax household income. Clearly, such differences between communities would have influenced the economic inequality we can observe from the tax lists. Also, in the absence of firm evidence of how incomes were taxed we cannot exclude that there was some variation in the scales adopted by these communities. Depending on the type of scale (proportional, regressive or progressive) similar levels of inequality statistics derived from fiscal sources may hide from view different realities of income distributions. For the Campine area, for example, the research of Eline Van Onacker has suggested that these communities adopted a more progressive tax scale. 53 The differences we could observe between communities in the index of fiscal dissociation also suggests that communities enjoyed significant autonomy and discretion in how they divided the fiscal burden in relation to the realities of the income distribution. In the absence of more evidence on local and regional variation in fiscal thresholds and scales we cannot exclude the possibility that fiscal sources from this period do not capture the realities of income distributions. A second factor that can explain these discrepancies stems from the metrics used to calculate inequality. Although these traditional metrics have the advantage that we can express inequality in a single measure that can be compared through time and space, measures such as the Gini index also may obscure some of the subtleties of income and wealth distributions. Similar Gini index values can correspond to different realities of distributions. More fine-grained measurement tools and techniques are required to expose these differences in wealth and income dis-52 W. RYCKBOSCH, Economic inequality and growth, cit. 53 E. VAN  tributions. 54 Thirdly, we must consider the possibility that traditional historical studies of rural societies overstated the impact of the underlying social and economic structures on economic inequality. In other words, in comparing regions in terms of the distribution of farm size, land ownership, common use rights and land, poverty and proletarianisation we may have focused too much on what sets them apart. More critical reflection and research on how these traditional social and economic parameters translated into wealth and income distributions is required.
Although at present our dataset raises more questions than it can provide answers, the results we obtained for Flanders and the Southern Low Countries clearly highlight some of the shortcomings and challenges of existing research. Our analysis, however, also offers a number of clues to guide future research on economic inequality in the past. Fiscal sources and data are undoubtedly valuable instruments to reconstruct economic inequality in the past, but also suffer from an important drawback. As this paper has shown, substantial local and regional variations can exist in how income was taxed. These local and regional differences in fiscal techniques, scales and technology can hide from view actual income distributions. As a result, cross-sectional comparisons of economic inequality based on such sources at the regional and international level are problematic. Also, in exploring the potential origins of the differences in inequality we observed in the countryside around Bruges we have shown that we lack a framework to evaluate and explain regional differences in economic inequality. 55 During the last years substantial progress has been made in identifying the long-term drivers of inequality. In particular, the nature and level of (state) taxation and proletarianisation have been identified as some of the key determinants of increasing economic inequality in late medieval and early modern Europe. 56 However, the factors that drove regional inequality in the longrun are probably different from those that shaped inequality at the local and regional level. In late medieval rural Flanders and the Southern Low Countries increased state taxation cannot explain regional differences in economic inequality as rural communities were exposed to the same level and method of state surplus extraction. 57 One possible way to advance research into the determinants and drivers of inequality suggested by this contribution, is to direct the focus to tracing changes over the long-term at the local level. The macro-economic theories that have been used by economic historians to explain long-term changes in economic inequality across pre-industrial periods do not seem adequate to account for the differences between 54 See for example G. ALFANI, M. DI TULLIO, The Lion's share, cit., chs 2 & 3; and the chapter by Stef ESPEEL and Sam GEENS in this volume. 55 See also the chapter by BAS VAN  regional patterns of inequality. Much theorizing has departed from general economic models such as Piketty's r > g hypothesis (which describes how inequality rises when the rate of return on capital is higher than economic growth), Williamson's skilled-unskilled wage rate (which measures the inequality produced by the skill premium), or the classic relationship of the land rent to the wage. 58 To a large extent such models derive from neo-classical economic theory that assumes the existence of functioning factor markets, and the commodification of labour, land and other means of production. However, as the case studies from fifteenth-century Flanders of exceptionally low inequality discussed above indicated, those were often linked to the particular organisation and regulation of economic resources such as fishing or common use rights in forests. It seems that a reliance on modern economic theory in discussing pre-industrial inequality has led to a neglect of precisely such institutions that were associated with only partial commofidication, as well as with lower levels of inequality. If we are to gain a better understanding of deviations in the trend towards growing inequality throughout early modern Europe, economic historians stand to gain much from adopting conceptual frameworks that can also take the political and social context of the local rural economy into account. A systematic comparison of long-term trends coupled with in-depth research on the social and economic fabric of rural communities has more potential to identify the factors that shaped and drove inequality than a more static and cross-sectional approach. In other words, to advance our understanding of the underlying forces resulting in high, medium or low inequality in rural settlements we need to adopt a more dynamic view that fully takes into account their specific production structures and their long-term evolution.