John McDonald FLINDERS UNIVERSITY

TAX FAIRNESS IN ELEVENTH CENTURY ENGLAND

Abstract: Alongside the Roman census from Augustus’ time and the ecclesiastical surveys or polyptychs of the 8th and 9th century Carolingian kingdoms, the Domesday Survey of 1086 occupies a most significant place in accounting history. Domesday Book, the outcome of the Survey, lists the incomes, tax assessments, wealth and resources of most estates in England and was used as a working accounting document by the monarch and public officials to raise taxes, distribute resources and consolidate power. Although the Domesday document itself survives, many details of its construction and use have been lost in the mists of time. This paper describes research to discover how taxes were levied and which estates and tenants received favorable treatment.

INTRODUCTION

In the accounting history literature, Godfrey and Hooper [1996] have convincingly argued that aspects of Domesday Book, the results of a survey commissioned by William the Conqueror, illustrate the concepts of accountability, decision-making and control.

Domesday Book served many purposes. It documented feudal tenancy arrangements and was a land register being used extensively to resolve land disputes in the courts. Indeed, the book’s name derives from this use. The manuscript refers to itself as the “Discriptio”, and it was only after Williams’ death referred to as “Domesday Book”, the book of last judgment, for

Acknowledgments: I would like to thank Eva Aker for excellent research assistance, Shawna Grosskopf, Rolf Färe and Knox Lovell for their support throughout the project, S. Snap and B. Moose for their caring assistance, one of the anonymous referees for very useful comments on an earlier draft of the paper, and The Australian Research Grants Council and Flinders University for financial support.

in land disputes there was no appeal beyond its pages — land rights could be traced to Domesday Book but no earlier.
As well as being a legal document, the book had a financial and decision-making purpose. It lists the incomes, tax assessments, wealth and resources of most estates in England and was used as a working accounting document by the monarch and public officials to raise taxes, distribute resources and consolidate power. As Godfrey and Hooper [1996, p. 51] state “By providing a valuation and audit of the resources of the feudal tenants-in-chief in 1086, Domesday enabled William and his successors to optimize both their wealth, through fiscal policy and efficient use of the country’s resources, and their power within the feudal structure of medieval England. For the English monarchy of the period, Domesday served both accountability and decision-making needs”.

Together with other ancient surveys that assisted financial accountability, the Roman tax census during the four centuries following Emperor Augustus and the ecclesiastical polyptychs of the 8th and 9th centuries which were used for tax and accounting purposes, the Domesday Survey occupies a landmark position in accounting history. Godfrey and Hooper [1996, p. 39] argue, “Domesday represents a partial extension of and evolution from what might be broadly termed public sector accounting as practiced in both the Roman and Carolingian periods”.

Although the Domesday document itself survives, many details of its construction and use have been lost. This paper describes research to discover how the taxes were levied and which estates and tenants received favorable treatment. Domesday Book records the tax assessments for the geld, a non-feudal tax levied by the king. The tax assessments are reported in hides and fiscal acres and are often referred to as the hidage system. In this paper, frontier methods are used to investigate who, and which estates, received beneficial hidation, and what factors were associated with favorable tax assessments.

DOMESDAY ENGLAND AND THE DOMESDAY SURVEY

The Domesday Survey was carried out 20 years after William invaded England from France. By 1086, Norman rule had been largely consolidated, although only after rebellion and civil dissent had been harshly put down. The Conquest was achieved by an elite. It did not result in a mass movement of people, and, although the Normans brought new institutions and practices, these were superimposed on the existing order. Most of the Anglo-Saxon aristocracy were eliminated, the lands of over 4,000 English lords passing to less than 200 Norman barons, with much of the land held by just a handful of magnates.

William I ruled forcibly through the Great Council. England was divided into shires, or counties, which were subdivided into hundreds. There was a sophisticated and long established shire administration. The sheriff was the king’s agent in the county, royal orders could be transmitted through the county and hundred courts, and an effective taxation collection system was in place.

England was a feudal state. All land belonged to the king. He appointed tenants-in-chief, both lay and ecclesiastical, who usually held land in return for providing a quota of fully equipped knights. The tenants-in-chief might then grant the land to sub-tenants in return for rents or services, or work the estate themselves through a bailiff.

Manorialism was a pervasive influence, although it existed in most parts of England in a modified form. On the manor the peasants worked the lord’s demesne in return for protection, housing, and the use of plots of land to cultivate their own crops. They were tied to the lord and the manor and provided a resident workforce. The demesne was also worked by slaves who were fed and housed by the lord.

Although Domesday Book records 112 boroughs, agricul-ture was the predominant economic activity, with stock rearing of greater importance in the south-west and arable faming more important in the east and midlands.

The Domesday Survey was commissioned on Christmas day, 1085, and it is generally thought that work on Domesday Book was terminated on the death of William in September 1087. The task was facilitated by the availability of Anglo-Saxon hidage lists. The counties of England were grouped into (probably) seven circuits. Each circuit was visited by a team of commissioners, bishops, lawyers and lay barons who had no material interests in the area. The commissioners were responsible for circulating a list of questions to land holders, for subjecting the responses to a review in the county court by the hundred juries, often consisting of half Englishmen and half Frenchmen, and for supervising the compilation of county and circuit returns. The circuit returns were then sent to the Exchequer in Winchester where they were summarized, edited and compiled into Great Domesday Book.

Unlike modern surveys, individual questionnaire responses were not treated confidentially but became public knowledge, being verified in the courts by landholders with local knowledge. In such circumstances, the opportunities for giving false or misleading evidence were limited.

Domesday Book consists of two volumes, Great (or Exchequer) Domesday and Little Domesday. Little Domesday is a detailed original survey return of circuit VII, Essex, Norfolk and Suffolk. Great Domesday is a summarized version of the other circuit returns sent to the King’s treasury in Winchester. (It is thought that the death of William occurred before Essex and East Anglia could be included in Great Domesday). The two volumes contain information on the net incomes (referred to as the annual values), tax assessments and resources of most manors in England in 1086, some information for 1066, and sometimes also for an intermediate year. The information was used to revise tax assessments and document the feudal structure, “who held what, and owed what, to whom”.1
The study described in this paper is based on data relating to 574 lay estates in the county of Essex in 1086. Essex was chosen because more detailed data are available on the counties described in Little Domesday, and the manorial entries for Essex are easier to interpret than those of Norfolk and Suffolk.2
1Further background information on Domesday England is contained in McDonald and Snooks [1986, Chs. 1 and 2; 1985a, 1985b, 1987a and 1987b] and McDonald [1998]. For more comprehensive accounts of the history of the period see Brown [1984], Clanchy [1983], Loyn [1962, 1965, 1983], Stenton [1943, 1951]. Other useful references includes Ballard [1906], Darby [1952], [1977], Galbraith [1961], Hollister [1965], Lennard [1959], Maitland [1897], Miller and Hatcher [1978], Postan [1966, 1972], Round [1895, 1903], the articles in Williams [1987] and references cited in McDonald and Snooks [1986]. The Survey is discussed in McDonald and Snooks [1986, sec. 2.2], the references cited there, and the articles in Williams [1987]. The Domesday and modern surveys are compared in McDonald and Snooks [1985c].

2 The data file was compiled by Eva Aker under the direction of the author with the aid of a Flinders University research grant. The file was compiled directly from Domesday Book entries in the Victoria County History of Essex which were checked against a facsimile of the Latin transcript and an English translation in the so-called Phillimore edition [Morris, 1975]. A general rule of thumb was developed that only entries for which (1) net income (annual value) is positive, (2) either ploughteams or livestock entries are positive (or both), and (3) there is a positive entry for at least one labour variable, were retained for analysis. In addition, seven other entries were deleted either because they were implausible or incomplete, and three others because no tax assessment was recorded. Further details are given in McDonald [1998].

EARLIER STUDIES OF THE GELD

The Domesday tax assessments relate to a non-feudal tax, the geld, thought to be levied annually by the end of William’s reign. The tax can be traced back to the danegeld, which was introduced by King Ethelred in 911 to provide finance to bribe or fight the Danes. Originally the geld was a land tax assessed at so much per hide. A hide was traditionally the acreage needed to support a man and his family, conventionally 120 acres, but in practice variable from place to place depending on the fertility of the land. Oldroyd [1997] describes the role of hidage lists and Geld Rolls in public accounting during the Anglo-Saxon period and their significance for accounting history. By Norman times it is thought that, although it retained the nomenclature of a land tax, the geld was no longer solely a tax on land. In 1086 it was one of a number of public revenue sources and probably contributed about a quarter of the total public purse. The geld was a significant impost on landholders, the rate struck in 1083-4 of six shillings to the hide, implies the tax amounted to about 15 percent of the annual value of the average Essex lay manor.

Domesday scholars have written extensively about the tax assessments. Much of the literature has been influenced by Round [1895], who considered the assessments to be “artificial”, in the sense that they were imposed from above via the county and hundred with little or no consideration of the capacity of an individual estate to pay the tax. Round’s view was largely based on a somewhat unsystematic and subjective review of the distribution of the assessments across estates, vills and the hundreds of counties.

In [1985a] and [1986, Ch. 4], Snooks and I argued that, contrary to Round’s hypothesis, the tax assessments were based on a capacity to pay principle, subject to some politically expedient tax concessions. Similar tax systems operate in most modern societies and reflect an attempt to collect revenue in a politically acceptable way.

There is empirical support for our hypothesis. Using re-gression methods, we showed, for example, that for Essex lay estates about 65 percent of variation in the tax assessments could be attributed to variations in manorial annual values (which measure the net income accruing to the lord) or mano-rial resources, two alternative ways of measuring capacity to pay. Similar results were obtained for other counties. Capacity to pay explains from 64 to 89 percent of variation in individual estate assessment data for the counties of Buckinghamshire, Cambridgeshire, Essex and Wiltshire, and from 72 to 81 percent for aggregate data for 29 counties [See McDonald and Snooks 1987a].

Although capacity to pay seems to explain most variation in tax assessments, some variation remains. Who was treated fa-vorably? Which estates received a beneficial hidation? And what factors were associated with beneficial hidation? Clearly, a first step in addressing these issues is to develop a measure of beneficial hidation.

A simple and appealing measure is based on the idea that an estate has received beneficial hidation if it has a lower tax assessment than another estate with the same or lower annual value (annual value or net income being a measure of capacity to pay). More formally, the beneficial hidation index (BHI) for estate i, is defined as the ratio of the maximum tax assessment of all estates with the same or a lower annual value than estate i, to the actual tax assessment of estate i. A BHI value of one corresponds to no beneficial hidation, and a value greater than one to some beneficial hidation.

JUSTIFICATION OF THE BHI

Some insight into the plausibility of the BHI (just defined) can be obtained by employing the frontier methodology (some-times used in production studies, see, for example, Lewin and Lovell, 1990). In Figure 1, A, B, C, D, and E indicate the tax assessments and annual values of five (fictitious) estates. (Estate A, for example, has an annual value of 5 shillings and tax assessment of 4 fiscal acres). To calculate the BHI for an estate, the maximum tax assessment for the estate’s annual value is required. The annual values of the five estates are 5, 10, 15, 20 and 30 shillings, and the maximum assessment for estates with equal or lower assessments 4, 10, 10, 30 and 30 fiscal acres, respectively. The maximum assessment values for different annual values can be thought of as describing a “tax frontier”.

The frontier that generates the BHI is illustrated in the up-per diagram of Figure 1. It consists of the “steps”, 0 to the point vertically below A, that point to A, the horizontal line from A to the point vertically below B, and so on. Estate BHIs are the ratio of maximum to the actual tax assessments. For estate E, the BHI=1.5, all other estates have a BHI=1. This frontier would be appropriate if the tax regime was one of constant tax assessment over annual value intervals (with, for example, estates with an annual value of 5 shillings and less than 10 shilling paying 4 fiscal acres; those with an annual value of 10 and less than 20 shillings, 10 fiscal acres, and so on), with some beneficial hidation.

Other tax frontiers and beneficial hidation indexes are plausible. For example, if the underlying tax regime consisted of multiple constant positive tax rate schedules (with, for example, estates with an annual value of 5 shillings and less than 10 shillings paying at one tax rate, those with an annual value of 10 and less than 20 shillings at a different rate, and so on), with some beneficial hidation, then the frontier is generated by starting at 0 and connecting points representing estates by line segments, so long as the slope of the segment is positive. This frontier is drawn in the lower diagram of Figure 1. 0 is connected to A, and A to B, because the line segments have positive slopes; but B is not connected to C and D not connected to E, because the slopes of the lines would not be positive (implying zero or negative marginal tax rates). Using this frontier, estates A, B and D have beneficial hidation indexes of one, the index for C is two, and for E, two and a half.

Unfortunately we do not know in detail how the Domesday tax assessments were formulated, so we do not know which is the most appropriate frontier, and hence beneficial hidation index.

It is reasonable to ask if it is possible, using empirical methods, to determine the “true” frontier. For example, is the true frontier the frontier that gives the closest fit to the data? Unfortunately, this may not be so. Casual inspection of Figure 1 indicates that the frontier in the upper diagram must always fit the data better than the frontier in the lower diagram (in the sense that the distances of the data points from the frontier cannot be greater and will sometimes be smaller), whether or not it is the true frontier (that is, whether or not the true tax regime is essentially one of constant tax assessment over annual value intervals).

In practice, if there are a reasonable number of observa-tions, well-distributed over the annual values, frontiers and in-dexes will be similar. The chosen frontier measures beneficial hidation more conservatively (in the sense that an estate’s index will tend to be smaller when measured against it) than most others. A major advantage in using it is that it can easily be calculated using linear programming methods [see, for example, McDonald, 1998, pp. 41-56].

BENEFICIAL HIDATION IN ESSEX IN 1086

When the frontier was constructed from the tax and annual value data for the 574 Essex lay estates in 1086, 18 estates lay on the tax frontier and so had a BHI=1.4 Figure 2 gives the frontier, the numbers on the frontier being the identification codes of the estates that form the frontier.
FIGURE 2 Tax Assessment Frontier. Essex Lay Estates, 1086

Table 1 gives the names of the estates and other information about them. For example, 1 refers to Fobbing an estate with an annual value of 720 shillings and a tax assessment of 2445.5 fiscal acres. All other estates are represented by points below the frontier. A few are located by a dot and their identification code, information about these estates being contained in Table 2.

4For any annual value, the frontier indicates the maximum tax assessment of all estates with that or a lower annual value, and an estate’s BHI is the ratio of the maximum assessment to the actual assessment of the estate.

TABLE 1 Characteristics of Estates with a Beneficial Hidation Index (BHI) of one. Essex Lay Estates, 1086

Estate BHI Tax Frontier Annual Tenant-in-chief Tenancy Hundred
assessment assessment value
1 Fobbing 1 2445.5 2445.5 720 Count Eustace Demesne Barstable
50 Tolleshunt Guines 1 1081 1081 110 Count Eustace 1 sub-tenant Thurstable
66 Elmdon 1 1680 1680 400 Count Eustace 1 sub-tenant Uttlesford
79 Lt. Bentley 1 42.5 42.5 3 Count Alan 1 sub-tenant Tendring
138 Wickford 1 1200 1200 180 Suen of Essex Demesne Barstable
219 Purleigh 1 840 840 100 Hugh de Montfort Demesne Dengie
250 Woodham Ferrers 1 1680 1680 560 Henry de Ferrariis Demesne Chelmsford
288 Stow Maries 1 637 637 65 Geoffrey de Magna Villa 1 sub-tenant Dengie
289 Saffron Walden 1 3120 3120 1000 Geoffrey de Magna Villa Demesne Uttlesford
293 Weneswic 1 640 640 80 Geoffrey de Magna Villa 1 sub-tenant Dengie
319 Wivenhoe 1 625 625 46 Robert Greno 1 sub-tenant Lexden
384 Debden 1 1980 1980 600 Ranulf Peverel Demesne Uttlesford
390 Down 1 1680 1680 260 Ranulf Peverel demesne Dengie
393 Stangate 1 1140 1140 160 Ranulf Peverel 1 sub-tenant Dengie
451 Ardleigh 1 195 195 17.67 Ranulf brother of Ilger 1 sub-tenant Tendring
481 Leyton 1 540 540 20 Robert son of Corbutio demesne Becontree
488 Paglesham 1 90 90 5 Robert son of Corbutio 1 sub-tenant Rochford
559 East Donyland 1 188 188 7 Ilbodo demesne Lexden
Note: Tax assessments are measured in fiscal acres and annual values in shillings.

TABLE 2 Characteristics of Selected Estates that Received Beneficial Hidation. Essex Lay Estates, 1086

Estate BHI Tax Frontier Annual Tenant-in-chief Tenancy Hundred
assessment assessment value
31 Boxted 2.00 600 1200 240 Count Eustace demesne Lexden
241 Stambrn/Toppesfld 9.88 170 1680 280 Hamo dapifer demesne Hinckford
374 Fairsted 15.27 55 840 100 Ranulf Peverel 1 sub-tenant Witham
247 Tiltey 18.02 60 1081 140 Henry de Ferrariis demesne Dunmow
115 How Hall 19.23 43.6 625 50 Richard son of C. Gilbert 1 sub-tenant Hinckford
453 Stevington End 25.44 42.5 1081 115 Tithel the Breton demesne Freshwell H-H
500 Sibil Hedingham 25.60 25 640 80 Roger Bigot 1 sub-tenant Hinckford
28 Toppes field 36.00 15 540 20 Count Eustace 1 sub-tenant Hinckford
207 Radwinter 36.00 15 540 30 Eudo dapifer 1 sub-tenant Freshwell H-H
555 Tendring 36.00 15 540 20 Moduin demesne Tendring
395 Prested 37.60 5 188 12 Ranulf Peverel 1 sub-tenant Lexden
195 Broxted 71.11 9 640 80 Eudo dapifer 1 sub-tenant Dunmow

Figure 3 exhibits the BHI histogram. Three percent of es-tates had a BHI= 1, about a quarter a BHI less than two, roughly a half an index value less than three, and three quarters a value less than five. Some estates had high BHI values. Seven percent had values of ten or more, with 195 Broxted and largest value of 71.11.

Table 1 provides summary information about the 18 estates lying on the frontier. No obvious patterns are evident for these estates. Some tenants-in-chief were major magnates, such as Count Eustace of Boulogne, Count Alan of Brittany, Suen, Sheriff of Essex and Geoffrey de Magna Villa, Sheriff of Middlesex, but several estates had tenants-in-chief who were less significant lords. In terms of tenancy, nine estates were held in demesne (that is, were worked by the tenant-in-chief) and nine had a single sub-tenant. Five of the frontier estates were in the hundred of Dengie, three in Uttlesford, two in Barstable, Lexden and

Tendring, and one in each of Rochford, Beacontree, Chelmsford and Thurstable. The estates seem to be well-distributed over the hundreds.

Turning to the estates with very high BHIs (greater than 18), from Table 2 it can be seen that they range from 395 Prested with a small annual value of only 12 shillings to a relatively large estate 247 Tiltey with an annual value of 140 shillings (140 shillings exceeds the annual value of more than three quarters of the estates in the sample). Of the nine estates with a BHI greater than 18, most had minor lords as tenants-in-chief, six were sub-tenancies and three held in demesne. Three estates were in Hinckford hundred, two in Dunmow, two in Freshwell half hundred and the others in Tendring and Lexden hundreds.

In footnotes to the Victoria County History entries for Essex [VCH, 1903], Round commented that four of the nine estates with very high BHIs had abnormal or nominal assessments. (These were 195 Broxted, 247 Tiltey, 28 Toppesfield and 500 Sibil Hedingham). He also commented on the low assessments of other estates with smaller BHIs.5 Round’s comments are rather unsystematic. By calculating BHIs for each estate it is possible to identify estates with low or abnormal assessments in a more comprehensive fashion.

STATISTICAL ANALYSIS OF FACTORS AFFECTING BENEFICIAL HIDATION

In the previous section the characteristics of estates with extreme BHI values were examined. Results of more compre-hensive analyses of factors associated with beneficial hidation are contained in Tables 3, 4 and 5.

Table 3 lists the mean BHI of estates of the 18 largest ten-ants-in-chief (those that had more than 10 estates in Essex). Eudo dapifer has the largest mean value (7.87). The deviation of this value from the overall mean (4.35) is 3.52. Notice, however, that the standard deviation of Eudo dapifer’s mean BHI is large (3.09). The high mean value is mainly due to the high BHIs of two of Eudo dapifer’s estates: 195 Broxted (BHI=71.11) and 207 Radwinter (BHI=36.00). Richard, son of Count Gilbert also has

5Examples are the assessments of 374 Fairsted (BHI=15.27) described as “strangely low” [VCH, 1903, footnote 4, p. 527], 571 Gestingthorp (BHI=8.00) also referred to as “strangely low” [footnote 9, p. 564], 241 Stambourne and Toppesfield (BHI=9.88) described as “an almost nominal amount” [footnote 4, p. 502] and 273 High Easter (BHI=3.26) “a very low hidation” [footnote 4, p. 509].

a high mean BHI (7.27), which is significantly greater than the overall mean.
Those who were not leniently treated include Robert, son of Corbutio (mean BHI=2.06), Robert Greno (mean BHI=2.73), Ralf Baignard (mean BHI=2.87), Ranulf, brother of Ilger (mean BHI=2.94) and Hugh de Montfort (mean BHI=2.97).

TABLE 3

Mean BHI of Estates of 18 Largest Tenants-in-chief. Essex Lay Estates, 1086

Tenant in chief Mean Standard Deviation Number of
BHI deviation from overall estates in
mean sample
Count Eustace 4.25 0.56 -0.10 71
Suen of Essex 4.19 0.48 -0.16 57
Geoffrey de Magna Villa 3.55 0.31 -0.80 42
Robert Greno 2.73 0.29 -1.62 44
Richard son of Count Gilbert 7.27 0.88 2.92 29
Ranulf Peverel 4.26 1.03 -0.09 37
Ralf Baignard 2.87 0.37 -1.48 29
Eudo dapifer 7.87 3.09 3.52 24
William de Warene 3.93 0.46 -0.42 18
Ranulf brother of Ilger 2.94 0.41 -1.41 17
Hugh de Montfort 2.97 0.47 -1.38 17
Hamo dapifer 4.33 0.66 -0.02 15
Peter de Valognes 4.76 1.31 0.41 14
Aubrey de Ver 4.76 0.82 0.41 16
Robert son of Corbutio 2.06 0.35 -2.29 11
Count Alan 4.50 1.13 0.15 9
Roger de Ramis 5.66 1.17 1.31 12
John son of Waleram 5.34 1.14 0.99 8
Others 4.77 0.51 0.42 104

There is a clear tendency for the tenants-in-chief with the largest number of estates in Essex to have less favorable assess-ments. 10 of the 12 largest tenants-in-chief have a mean BHI below the overall mean (4.35), and all but one of the remaining six tenants-in-chief a mean above the overall mean. The vast majority of tenants-in-chief fall in the ‘other’ category. Their mean BHI is also above the overall mean, indicating that they tended to be treated more leniently.

A more objective way of assessing whether, in general, ten-ants-in-chief were treated equally is to carry out a statistical test using the full sample of observations. A robust statistical test of

the null hypothesis that the mean BHIs for the tenants-in-chief are equal, resulted in rejection of the null at the five and one percent significance levels.6 The test indicates that who the ten-ant-in-chief was is a significant factor influencing how estates were taxed, with some, mainly smaller, tenants-in-chief receiv-ing more favorable treatment than others.

Figure 4 is a map indicating the Essex hundred divisions and Table 4 gives a breakdown of mean BHI by hundreds. A statistical test indicates that the BHI varied significantly (at the five and one percent levels) with hundred location.7 Hundreds

6 The test was carried out by regressing the BHI on tenant-in-chief dummy variables taking the value 1, if the tenant-in-chief held the estate; 0, otherwise. Since the regression diagnostics indicated heteroskedasticity in the disturbances, White’s [1980] heteroskedasticity-consistent test was used. On the null, the test statistic is asymptotically distributed as a F-distribution with 18 and 555 degrees of freedom. The test statistic value was 4.293 which, to five decimal places, has a p-value of zero.

7The test was carried out in a similar way to the test for equality of the tenant-in-chief means (using White’s method, see previous footnote). The test statistic value (asymptotically F-distributed with 21 and 552 degrees of freedom on the null) was 11.085, which, to five decimal places, gives a p-value of zero.

for which estates received milder assessments included Fresh-well half hundred (mean BHI = 8.94), Hinckford (mean BHI = 7.43), Dunmow (mean BHI = 6.60), Lexden (mean BHI=5.57) and Maldon half hundred (mean BHI=5.40). Those less-well treated were Beacontree (mean BHI=1.55), Dengie (mean BHI=2.34), Clavering hundred and half hundred (mean BHI = 2.37), Winstree (mean BHI = 2.38), Waltham (mean BHI=2.39), Chafford (mean BHI=2.43), Chelmsford (mean BHI=2.52) and Barstable (mean BHI=2.55).

TABLE 4 Mean BHI of Estates by Hundred. Essex Lay Estates, 1086

Tenant in chief Mean Standard Deviation Number of
BHI deviation from overall estates in
of mean mean sample
Barstable 2.55 0.21 -1.80 35
Beacontree 1.55 0.10 -2.80 9
Chafford 2.43 0.39 -1.92 12
Chelmsford 2.52 0.14 -1.83 48
Dengie 2.34 0.25 -2.01 41
Dunmow 6.60 1.44 2.25 48
Clavering hundred and
half hundred 2.37 0.33 -1.98 10
Freshwell half hundred 8.94 2.18 4.59 17
Harlow 3.61 0.88 -0.74 18
Harlow half hundred 2.97 0.77 -1.38 3
Hinckford 7.43 0.67 3.08 73
Lexden 5.57 1.21 1.22 31
Ongar 5.09 0.51 0.74 34
Rochford 4.10 0.57 -0.25 36
Tendring 3.74 0.74 -0.61 48
Uttlesford 2.90 0.27 -1.45 39
Waltham 2.39 0.80 -1.96 4
Winstree 2.38 0.26 -1.97 15
Witham 4.83 0.75 0.48 26
Maldon half hundred 5.40 3.60 1.05 2
Thunreslau half hundred 3.30 0.72 -1.05 3
Thurstable 2.89 0.36 -1.46 22

The above analysis indicates that all estates were not treated equally, but that tax treatment varied significantly across ten-ants-in-chief and the hundreds. An obvious question to ask is, if, when we allow for the hundred effect, the tenant-in-chief effect

is still significant, and, if, when we allow for the tenant-in-chief effect the hundred effect remains significant. Extending the argument we could examine the relationship between the BHI and all factors that might plausibly be expected to affect it and for which information is available at the estate level. Multiple regression could then be used to estimate the relationship and test whether one factor (for example, who the tenant-in-chief was) significantly affects the index when all other factors are controlled for.

This approach was implemented. As well as who the tenant-in-chief was and hundred location, information is available, estate by estate, on whether the estate was close to an urban center, the size of the estate, the kind of agriculture practiced and the tenure arrangement on the estate, all factors that could affect an estate’s tax assessment. Table 5 exhibits the main results of a regression of the BHI on variables measuring these characteristics.8 Details of the implementation of the hidage system are now largely unknown, so the regression will provide empirical evidence as to whether particular groups or activities received special treatment, and, given these special considerations, whether the assessments were evenly distributed over the county.

The results show that the tenant-in-chief and hundred ef-fects remain significant when other factors are allowed to vary in the multiple regression. Whether the estate was close to Maldon or Colchester was also a significant factor. The BHI for

8 Tenant-in-chief was indicated by 18 dummy variables (the ith, i=1 … 18, taking the value 1, if the ith largest tenant-in-chief held the estate; 0, otherwise; the intercept measuring the effect when none of the 18 largest tenants-in-chief held the estate), and the hundred location by 21 dummies (with the intercept measuring the effect of location in Thurstable hundred). Colchester and Maldon were the main towns in Essex. The effect of proximity to an urban centre was measured by a dummy variable, taking the value 1, if the estate was in an approximate six mile radius of Colchester or Maldon (allowing for topology); 0, otherwise. Size was measured by the single best indicator of the economic size of an estate, the estate’s annual value. An index of whether production was mainly arable or grazing is given by the grazing/arable ratio, defined as livestock less cattle and beasts (which were required for ploughing) divided by the number of ploughteams on the estate. (Livestock less cattle and beasts is a weighted average of swine, sheep and goats with prices as weights. Three estates had no ploughteams. For them, the ratio was set at 2000, the largest ratio value for estates with some ploughteams being 1376). Finally, tenure was measured by dummy variable taking the value 1, if the estate was held in demesne; 0, otherwise. Test statistics are heteroskedasticity-consistent tests statistics obtained by White’s [1980] method.

TABLE 5
Regression of BHI on Estate Characteristics. Essex Lay Estates, 1086

Test statistic Distribution P-value
on null
Tenant-in-chief effect 1.857* F(18,530) .017
Hundred effect 5.164** F(21,530) .000
Urban centre effect -3.1** t(530) .002
Size (annual value) effect -4.0** t(530) .000
Kind of agriculture (grazing/
arable ratio) effect -1.1 t(530) .255
Tenure effect -2.2* t(530) .028

Note: The tests are heteroskedasticity-consistent tests [see White, 1980].
* indicates significant at the five percent level and ** significant at the one percent level. R2=.17

estates close to these towns was, on average, 1.73 lower than for other estates. Economic size (measured by annual value) of the estate also significantly affected the index value. A large estate (with an annual value of 320 shillings) had an average index value 1.80 less than a small estate (with an annual value of 20 shillings). Whether or not an estate was held in demesne was a significant factor at the five percent level. Estates held in demesne, on average, had a BHI 0.91 less than those that were sub or mesne-tenancies. The variable measuring the mix of arable and grazing agriculture on an estate was not a significant correlate.

CONCLUSION

The paper has presented the results of an investigation into the incidence of favorable tax assessment (hidation) in Domesday Essex. Frontier methods were used to derive a measure of beneficial hidation, and estates with favorable and unfavorable assessments identified. Tenants-in-chief and local areas (hundreds) of the county with lenient assessments were discovered, and regression methods used to assess the sig-nificance of the association of characteristics of estates and beneficial hidation. Factors significantly associated with beneficial hidation were the tenant-in-chief holding the estate (hidation tended to be less beneficial for the tenants-in-chief holding a large number of estates in Essex), the hundred location, proximity to an urban center (estates remote from the urban centers being more favorably treated), economic size of the es-tate (larger estates being less favorably treated) and tenure (es-tates held as sub-tenancies having more lenient assessments). The kind of farming undertaken (arable or grazing) was not a significant factor.

The details of the levying of the geld in 1086 are largely lost in time, but the evidence clearly indicates that the manorial tax assessments were based on a capacity to pay principle (as measured by the manor’s annual value), and the analysis of estate BHIs shows that other factors also had a significant influence.
In most tax systems, certain groups or activities receive concessions and the administrative process induces unevenness in the assessments. The BHI analysis indicates that, allowing for the capacity of an estate to meet the tax, some estates were indeed favored above others. The results show that some tenants-in-chief were treated more leniently than others, and, interestingly, it tended to be the tenants-in-chief holding fewer rather than more estates in the county. At the margin, the assessment system may have tended to favor the less wealthy because, it was also found that smaller estates and those held by sub-tenants received lower assessments, and urban estates (often held by the wealthy), higher assessments.

The fact that there was a significant hundred assessment differential, suggests that administrative factors affected the hidage system. This could have been because the assessments were made at different dates or with (slightly) more rigor in some hundreds than others. As for concessions being given when particular activities were undertaken, the regression pro-vides no evidence of this. In particular, the tax system did not favor arable activity over animal husbandry or vice versa.

REFERENCES

Ballard, A. (1906), The Domesday Inquest (London: Methuen).
Brown, R.A. (1984), The Normans (Woodbridge: Boydell Press).
Clanchy, M.T. (1983), England and its Rulers, 1066-1272 (Glasgow: Fontana).
Darby, H.C. (1952), The Domesday Geography of Eastern England, reprinted 1971 (Cambridge: Cambridge University Press).
Darby, H.C. (1977), Domesday England, reprinted 1979 (Cambridge: Cambridge University Press).
Darby, H.C. and I.S. Maxwell (eds) (1962), The Domesday Geography of Northern England, (Cambridge: Cambridge University Press).
Galbraith, V.H. (1961), The Making of Domesday Book (Oxford:Clarendon Press).
Godfrey, A. and Hooper, K. (1996), “Accountability and Decision-making in Feu-dal England: Domesday Book Revisited,” Accounting History, NS Vol. 1, No. 1: 35-54.
Hallam, H.E. (1981), Rural England 1066-1348 (Brighton: Fontana).
Hollister, C.W. (1965), The Military Organisation of Norman England, (Oxford:
Clarendon Press). Langdon, J. (1982), “The Economics of Horses and Oxen in Medieval England,”
Agricultural History Review, Vol. 30, No. 1: 31-40. Lennard, R. (1959), Rural England 1086-1135: A Study of Social and Agrarian
Conditions (Oxford: Clarendon Press). Lewin, A.Y. and C.A.K. Lovell (eds) (1990), “Frontier Analysis: Parametric and
Nonparametric Approaches,” Journal of Econometrics, Vol. 46, No.1: 1-245. Loyn, R. (1962), Anglo-Saxon England and the Norman Conquest, reprinted 1981
(London: Longman).
Loyn, R. (1965), The Norman Conquest, (London: Hillary House). Loyn, R. (1983), The Governance of Anglo-Saxon England, 500-1087 (London:
Edward Arnold). McDonald, J. (1997), “Manorial Efficiency in Domesday England,” Journal of
Productivity Analysis, Vol. 8, No.3: 199-213. McDonald, J. (1998), Production Efficiency in Domesday England (London:
Routledge). McDonald, J. (2000), “Domesday Economy: An Analysis of the English Economy
Early in the Second Millennium,” National Institute Economic Review, Vol.
172, No.2: 105-114. McDonald, J. and G.D. Snooks (1985a), “Were the Tax Assessments of Domesday
England Artificial? The Case of Essex,” Economic History Review, Vol. 38,
No. 4: 353-373. McDonald, J. and G.D. Snooks, (1985b) “The Determinants of Manorial Income
in Domesday England: Evidence from Essex,” Journal of Economic History,
Vol. 45, No.4: 541-556. McDonald, J. and G.D. Snooks (1985c), “Statistical Analysis of Domesday Book
(1086),” Journal of the Royal Statistical Society, Series A, Vol. 148, No.2: 147-
160. McDonald, J. and G.D. Snooks (1986), Domesday Economy: A New Approach to
Anglo-Norman History (Oxford: Clarendon Press). McDonald, J. and G.D. Snooks (1987a), “The Suitability of Domesday Book for
Cliometric Analysis,” Economic History Review, Vol. 40, No.2: 252-261. McDonald J. and G.D. Snooks (1987b), “The Economics of Domesday England,”
in Domesday Book Studies, A. Williams (ed.) (London: Alecto Historical Editions): 86-89. Maitland, F.W. (1897), Domesday Book and Beyond, reprinted 1921,
(Cambridge:Cambridge University Press). Miller, E., and J. Hatcher, (1978), Medieval England: Rural Society and Economic
Change 1086-1348 (London: Longman). Morris, J., (gen. ed.) (1975), Domesday Book: A Survey of the Counties of England
(Chichester: Phillimore).
Oldroyd, D. (1997) “Accounting in Anglo-Saxon England: Context and Evidence,” Accounting History, NS Vol. 2, No.1: 7-34. Postan, M.M. (1966), “Medieval Agrarian Society in its Prime”, The Cambridge
Economic Historyof Europe, Vol. 1, M.M. Postan (ed.) (Cambridge: Cambridge University Press): 41-79. Postan, M.M. (1972), The Medieval Economy and Society: An Economic History of
Britain in the Middle Ages (London: Weidenfeld & Nicolson). Raftis, J.A. (1957), The Estates of Ramsey Abbey: A Study in Economic Growth
and Organisation (Toronto: Pontificial Institute of Medeival Studies).
Round, J.H. (1895), Feudal England: Historical Studies on the Eleventh and
Twelfth Centuries, reprinted 1964, (London: Allen & Unwin). Round, J.H. (1903), Essex Survey, in VCH Essex, Vol. 1, reprinted 1977 (London:
Dawson). Stenton, D.M. (1951), English Society in the Early Middle Ages, reprinted 1983,
(Harmondsworth: Penguin).
Victoria County History (1900) (London: Oxford University Press). White, H. (1980), “A Heteroskedasticity-consistent covarience matrix estimator
and a direct test for heteroskedasticity,” Econometrica, Vol. 48, No.5: 817-
838. Williams, A. (ed) (1987), Domesday Book Studies (London: Alecto Historical
Editions).