Introduction to the Tafel v-bis Dataset: Death Duty Summary Information for The Netherlands, 1921 Social and Economic History

This article introduces a newly constructed dataset (i.e. the Tafel v-bis Dataset) containing summary information for all Dutch citizens who died in 1921 and were subject to inheritance taxation. This dataset provides personal and socio-economic information on 24,263 individuals, including their total wealth, age, profession, residence and marital status at their time of death. Consequently, this dataset can be useful for researchers stemming from various academic disciplines. The article first discusses the range of possible uses of the dataset. Then the authors explain how they constructed the dataset and provide the necessary criticism regarding the underlying source mate-rial. The dataset is available via the Utrecht University data platform, yoda.


Introduction
The goal of this article is to introduce a newly constructed dataset (i.e. the Tafel V-bis Dataset) containing summary information for the entirety of the population of Dutch citizens who died in 1921 and were subject to inheritance taxation. This dataset has personal and socio-economic information on 24,263 individuals, which includes their total wealth, age, profession, residence and marital status at their time of death. Consequently, this dataset can be useful for researchers stemming from various academic disciplines. Sociologists and demographists with interest in history, for example, can explore the correlation between an individual's gender, their profession and their accumulated wealth over a lifetime. They could also trace the root of regional development and gentrification, or the deterioration of regions on the municipal level. Since there are very few sources that combine information about wealth and social variables, this dataset could help us better understand how social stratification functioned and where social classification systems such as Hisco or Hisclass (see Appendices A and B) can be improved. Genealogists, on the other hand, could find personal information on their ancestors which is unavailable elsewhere. Legal scholars can rely on this data and their underlying sources to gain a deeper understanding of notarial deeds and the system of inheritances. Finally and perhaps most fittingly, (economic) historians can use this data to draw a very detailed picture of the distribution of wealth in the Netherlands in the early twentieth century (Bos, 1989;de Vries, 1986;Moes, 2012;van der Burg & Rhoen, 2005;Wilterdink, 2015).
The interest in inequality has recently increased. Most notably are the works by Piketty (2014) and Piketty et al. (2006Piketty et al. ( , 2014 about wealth inequality in the Western world and Paris in particular. Nonetheless, the availability of data, especially historical data, has not appeared to follow suit. Moreover, very few individual-level data are available on a national scale. We partially fill this gap for the Netherlands by constructing a database based on a source that brings together both financial and non-financial information on an individual level. To do this, we have selected the memories van successie or death duty forms which result from the levying of death duties on estates. These were customary in some of the Dutch Republic's provinces from the late seventeenth century onward and nationwide from 1818 to 1927 (Bos, 1989). These series are continuous and publicly accessible in the Dutch regional archives, and are readily available on paper, microfilm or online.
This article proceeds as follows. In section two we provide a detailed description of the dataset. We first describe the process through which the original source was created and apply source critique to the administrative procedure, particularly paying attention to the correctness of the wealth variables. Then we explain how we transcribed the source and how we constructed the dataset and give a brief explanation of how the extra wealth class variable was created.

2.
Description of the Dataset -Tafel V-bis Dataset deposited at the Utrecht University data platform yoda -doi:www.doi.org/10.24416/UU01-QG9Q8B -Temporal coverage: 1921 (1 January -31 December) The Tafel V-bis Dataset contains the information on all persons who died between 1 January 1921 and 31 December 1921 in the Netherlands and who were subject to inheritance taxation. This was about one-third of the ca. 66,000 deceased of 1921(Centraal Bureau voor de Statistiek, 1921. This source, bureaucratically known as Tafel V-bis (Table Five-bis) is, in fact, a ledger containing references to and summary information of the underlying death duties, in Dutch called memories van successie. The memories formed the basis of the Dutch inheritance taxation system between 1818 and 1927. The Tafel V-bis exists as a complementary source to the memories since 1856. The Tafel V-bis is organized in a ledger per tax office (kantoor) and lists every individual for whom a memorie was drawn up in, organized per letter, and a summary description of personal information. The ledgers are currently held in the respective regional branches of the Dutch National Archive. Our database consists of a transcription of all ledgers of death duties for 1921, not the individual underlying death duties.
We have chosen the year 1921 because it is the year following the national census of 1920 (dated 31 December 1920). The census data gives us a clear overview of the country, its population, age distribution and employment on a local level. Unfortunately, since individual-level census information is no longer available, we cannot connect both sources. Nonetheless, the overview on a more aggregate level does tell us more about specific biases in our dataset. The most notable bias by far is the age bias. The individuals in our dataset are notably older than the general Dutch population, and slightly older than the people who passed away in the Netherlands in 1921.
Moreover, the individuals in the dataset represent the right tail of the national wealth distribution. Figure 2 visualizes the distributions. We only capture the wealthiest 20% of the group of deceased. The dotted line represents the 1,000 guilder cutoff point. The Gini-coefficient for the Tafel V-bis dataset is 0.852.

Creation of the Source
For a good understanding of the dataset and a critical assessment of its possible applications, we must pay attention to the administrative process that created the underlying source. In particular, our accuracy of the wealth variable depends on the quality of the underlying memorie.
The format of a memorie is similar to the more frequently researched probate inventories (boedelinventaris) in that each memorie mentions the name of deceased and the heirs followed by a description of the composition and value of the inheritance, bequest or gift. In addition, it mentions the profession and place of birth and residence of the deceased and the degree of kinship between the deceased and their respective heirs. Further, it mentions whether there was a will or prenuptial agreement with the name and residence of the notary who underwrote the agreement and the date when this took place. Considering both have a clear impact on how the inheritance ought to be divided, this information is of great importance. Each memorie provides a structured description of the inheritance and an estimate of the value of each individual asset. In the case of real estate, the nature, size, location, and cadastral registration numbers are also listed. Finally, each memorie also included a detailed account of the outstanding debt and accounts receivable of the deceased and the name of the counterparty.
The difference between memories and probate inventories is that memories were drawn up with the clear goal of taxing the estate, whereas probate inventories were compiled to, for example, safeguard the inheritance of minors or when people married (Wijsenbeek-Olthuis, 1995). This makes that, with memories, it was in the interest of the heirs not to register the whole inheritance to (illegally) lower taxes. The procedure, however, seems to have been relatively strict, leaving little room for such tax evasion.
Inheritances were only registered when they were subject to levies. Before 1878, only estates with a net value larger than 300 guilders and when not all heirs were descendants in the direct line, were subject to the duty. Afterwards, every estate with a net value over 1,000 guilders had to pay the duty (Nationaal Archief, 2019). The broadening to all estates over a threshold, regardless of the relation of the heirs, greatly expanded the number of taxable estates. In 1921, every estate larger than 1,000 guilders was subject to the tax. Even corrected for inflation, the threshold went up. 300 guilders before 1878 consistently equalled between 2,000 and 3,000 2018 euro, whereas 1,000 guilders in 1878 is equal to roughly 10,000 2018 euro. Strong inflationary pressure during the First World War lowered the effective threshold for taxation, increasing the percentage of the population subject to taxation: 1,000 1921 guilders equals ca. 6,500 2018 euro (Internationaal Instituut voor Sociale Geschiedenis, 2020).
Initially, Death Duties were drawn up by designated officials, usually municipal secretaries, known in Dutch as Secretaris van het plaatselijk bestuur, or Gequalificeerde tot de directie der invordering van de Belasting op het Regt van Successie. They were responsible for verifying the declarations made by families of the deceased at the house of the deceased -or sterfhuis in Dutch. Over time this procedure, which proved too much of an administrative burden, changed several times. From the 1900s onwards, representatives of the deceased (e.g. heirs, legatees, custodians, testamentary executors) were obliged to present a declarative statement to the municipal and/or cantonal office of the last place of residence of the deceased. This declaration had to be in written form and had to give insight into the nature and value of the inheritance. Everyone, including institutions and individuals exempted from taxation, had to file such a declaration (Zeeuws Archief, 2018). In case of the inheritance being worth less than 300 guilders, the heirs had to hand in a Certificate of 7 the Tafel V-BIS DATASET | 10.1163/24523666-bja10007 research data journal for the humanities and social sciences (2020) 1-19 Inability (Certificaat van Onvermogen), which they obtained from the municipality where the deceased resided.
This declaration had to be drawn up within six months after the time of death. Following this declaration, the heirs had one month (before 1911, two weeks) to take an oath in front of the District Court (Arrondissementsrechtbank) or the Subdistrict Court Judge (Kantonrechter), to certify that the provided information was correct. This official report (proces-verbaal) of the oathtaking had to be presented within four weeks to the civil servants in charge of the Death Duty and was attached to the declaration. Following this pledge, civil servants verified the declaration by cross-referencing the names and numbers of the deceased -based on a variety of sources, such as wealth and income taxes, at their disposal -within a given municipality or canton, with the individual names and the total number of declarations (Het Utrechts Archief, 2020). In the transitory period of six weeks, the corresponding levy was determined by the appropriate officials. Once the heirs were made aware of the amount due, they were given six weeks to resolve the outstanding levy.
Heirs were allowed to provide an additional supplementary declaration, socalled suppletoire aangifte, without any penalties in case of a further amendment to the initial declaration of inheritance. Usually, this happened at the insistence of the authorities. In case this was insufficient, or if the heirs did not cooperate, the authorities could go and estimate the value of the assets themselves. In case any fraud or tax evasion was detected, the authorities ordered a professional estimate of the inheritance and court proceedings could be started. Penalties were more or less twice the due tax plus costs of estimation (Zeeuws Archief, 2018). The effectiveness of these penalties was, of course, subject to the enforceability and ability to verify. Real estate was hard to hide and its value was easily verifiable using the cadaster, but the values of especially foreign and unlisted shares were much harder to establish. We can expect most fraud and tax-evasion to have taken place in these asset categories (Zeeuws Archief, 2018).

Transcription of the Source
A team of research assistants meticulously transcribed the Tafel V-bis of 1921 for the entire Netherlands.1 The following table gives an overview of all the variables in the dataset with their names, explanation and name in the original source. Every variable that was not in the original source is marked by an asterisk behind the name: 8

Ref_Reg
Reference number to the underlying memorie van succession. The original source has "/" instead of "#". All "/" were replaced to prevent import/ conversion errors.
2. Doorlopend volgnummer van het register no. 4 Family_ Name Family Name of the deceased 3. Naam First_Name First Name of the deceased 4. Voornamen Gender* 0=male, 1=female, GO=unknown. These were added on the basis of first names combined with information from the "Wie was wie" database. [a] Whenever it was unclear GO was assigned. This process was fully manual.
Profession Profession of the deceased as listed by the heirs. Most likely this is the occupation at death.

Beroep
Hisco* Hisco classification of the profession. This was manually assigned using the 2018 Hisco code book. [b] Hiscam* Historical Camsis classification, based on Hisco. This was automatically converted to Hiscam L: Later period. c1890-1938, using a conversion script.   Originally, the variables Day_Birth, Month_Birth, Year_Birth, and Place_Birth were listed in one entry in the source (column 9). For practical reasons, we have split this information into four variables. The variables indicating divorce and children were also mentioned in the source as part of the variable 12 on marital status (column 8). We have split this information and coded it into dummy variables, again for practical reasons.
To keep track of ambiguities in the source and problems with the transcription of the source itself, such as illegible or confusing entries, the research assistants have noted this in the variables Info_Logbook and Sum_Logbook. Afterwards, we converted these comments into the dummy variables VALIDnetvalue, VALIDname&profession and VALIDAge to indicate the quality of the entries.
The dataset is not a complete transcription of the source. We have not transcribed column 1 and columns 11 to 21, which refer to other actions and documents. We were not interested in these actions and have not transcribed them. Also, only occasionally information was entered in these columns. We have photographs of them, which can be made available on request so that other researchers can expand the dataset if they need this information.

Construction of the Wealth Class Variable
To get a better insight into the data and allow for more meaningful comparisons, we constructed six different wealth classes based on the provincial wealth distributions derived from the source (i.e. the Tafel V-bis). We discerned the following wealth classes: 1. Those who own less than 1000 f.: This class contains everyone that did not meet the official registration cut-off point. Probably these are people with high total assets, but also a lot of debts. We excluded them from the group for which we calculated the classes. Besides, we have categorized the remaining population by their % of wealth: Comments and references to the relative memoria and articles in case of prosecution in relation to reporting, etc.

14
The barriers to class are set per province and are thus relative to the provincial wealth distribution. The average cut off point for belonging to the provincial 1% is 328,796.55 guilders. The lowest entry into the 1% class was in Limburg, where 83,052 guilders sufficed. The highest entry point was in Utrecht, where you needed to own at least 742,281 guilders worth of assets. In Figure 2 above, we have provided a graphical overview of these cut-off points. These numbers should not be compared with national estimates, because our threshold for belonging to the regional 1% is considerably higher than belonging to the national 1%. It has to do with the inherent bias of only looking at people who pay inheritance taxes. The cut-off point for belonging to the national 1% in 1919 was around 80,000 guilders. When accounting for the number of people who are not in our dataset, we come to cut-off points comparable to the national ones. Counting the wealthiest 1% of the total Tafel V-bis data set, we arrive at a cut-off point of 108,945 guilders when not counting infants and minors1 and 87,107 guilders when counting everyone who passed away.2

Concluding Remarks
This dataset is a first step in tapping this long-term source of high-quality information on wealth distributions linked to social variables. We are currently working with the Clariah project to link this dataset to civil registers and bring more data together on a personal level. We hope that this article will show the possibilities of this dataset and inspire other researchers to extend it to other years.
the Tafel V-BIS DATASET | 10.1163/24523666-bja10007 research data journal for the humanities and social sciences (2020) 1-19 Appendix A Table A1 hisco Tree of Occupational Groups [a] Major group 0/1 Professional, technical and related workers Workers in this major group conduct research and apply scientific knowledge to the solution of a variety of technological, economic, social and industrial problems and perform other professional, technical, artistic and related functions in such fields as the physical and natural sciences, engineering, law, medicine, religion, education, literature, art, entertainment and sport.

Major group 2 Administrative and managerial workers
Workers in this major group conduct research and apply scientific knowledge to the solution of a variety of technological, economic, social and industrial problems and perform other professional, technical, artistic and related functions in such fields as the physical and natural sciences, engineering, law, medicine, religion, education, literature, art, entertainment and sport.

Major group 3 Clerical and related workers
Workers in this major group put into effect laws, rules and regulations made by central, state, provincial or local governments; supervise clerical and related work, transport and communications service operations; compile and maintain records of financial and other business transactions; handle cash on behalf of an organisation and its customers; record oral or written matter by shorthand writing, typing and other means; operate office machines and telephone and telegraph equipment; conduct passenger transport vehicles; take part in postal work and mail distribution and perform other duties related to the foregoing.

Major group 4 Sales workers
Workers in this major group are engaged in, or directly associated with, buying and selling goods and services of all kinds and in conducting wholesale and retail businesses on their behalf.

Major group 5 Service workers
Workers in this major group organise or perform catering, housekeeping, personal, protective and related services.

Major group 6 Agricultural, animal husbandry and forestry workers, fishermen and hunters
Workers in this major group conduct farms on their behalf or in partnership, perform agricultural, animal husbandry and forestry tasks, catch fish, hunt and trap animals, and perform related tasks.

Major group 7-8-9 Production and related workers, transport equipment operators and labourers
Workers in this major group are engaged in or directly associated with the extraction of minerals, petroleum and natural gas from the earth and their treatment; manufacturing processes; the construction, maintenance and repair of various types of roads, structures, machines and other products. Also included are those who handle materials, operate transport and other equipment and perform labouring tasks requiring primarily physical effort.

Appendix B
hisclass hisclass is international historical class scheme, created for the purpose of making comparisons across different periods, countries and languages. Furthermore, it is linked to an international standard classification scheme for occupations -hisco. hisclass is an instrument that can be used to systematically compare social class positions, distilled from a dazzling variety of occupational titles, around the world and over a range of periods.3 Table A2 Description of Hisclass [a] hisclass Description 1 Higher managers 2 Higher professionals 3 Lower managers 4 Lower professionals, clericals and salesmen 5 Lower clericals and salesmen 6 Foremen 7 Skilled workers 8 Farmers 9 Lower skilled workers 10 Lower skilled farm workers 11 Unskilled workers 12 Unskilled farm workers [a] Mandemakers et al., 2018.