Did the state take little? What peasants handed over, read from the original registers

Opening

In 1844–45 the Ottoman Empire carried out an empire-wide survey of income. In every village a clerk wrote down each household's income item by item, and on the same page he recorded how much direct tax the household had paid the previous year (vergi-i mahsusa, a tax apportioned by household) and how much tithe it paid that year. For 17 villages of Kepsut district, in today's Balıkesir province in Turkey, a researcher has transcribed these registers household by household. Adding up 290 households, these two payments to the state came to 25.1 per cent of all registered income: roughly one part in four went out.

For the same empire, economic historians' aggregate estimates put the central treasury's revenue at only 2 to 3 per cent of gross domestic product at the end of the eighteenth century, and only a little over a tenth on the eve of the First World War (Pamuk 2009). The two figures differ so much because, between the peasant and the treasury, taxes passed through tax farmers, local notables and officials at every level, and the treasury received only a small part. So whether "the state took little" depends on which end is measured. Looked at from the treasury, the burden was light. Looked at from what the registers record under the peasants' names, it was not.

This page takes the original registers that could be found and computes that share for the whole population of a place: a village, a district, a sub-province or a province. The Ottoman material is the richest; Russia and Japan have a few places each; for England and Qing China the gaps are filled for now with figures from secondary studies that name a place and a date.

Chart: shares handed over, place by place

Each point is what the whole population of one place handed over, as a share of its total registered income or output. Colour shows who received it: the state's direct tax; the state's direct tax plus the tithe; a lord or landlord; the church; or a combined total that cannot be split. Each point is labelled with the place and year. Where a source gives a range, it is drawn as a line.

Ottoman Empire: places within one empire differed several times over

More than twenty villages, towns and districts from the 1844–45 registers have now been transcribed or computed. Counting only the direct tax apportioned by household, most places fall between 11 and 24 per cent; adding the tithe, most fall between 18 and 30 per cent. The tithe at this time went to the state, which farmed it out to tax farmers, so it is counted as part of the state's share.

The lowest places are at a few per cent: in Tivnik village near Erzurum in the east, direct tax plus tithe came to 9.2 per cent. The highest are above 30 per cent: Saruhanlar village in Kütahya reached 37.8 per cent, and so did one village in Kilmigad district.

In 1847 the Ottoman central government itself compared the registers sent in from each sub-province (sancak) to judge where the burden was heavy and where light. By sub-province, the 1845 direct tax as a share of registered income was only 5 per cent in Varna and 5.5 per cent in Silistra in the Balkans, but 31.5 per cent in Ankara and 37 per cent in Kastamonu in Anatolia. On this basis the centre judged some places overburdened and others underburdened, and set out to redistribute the tax between sub-provinces. The central government of the time was itself aware of how far the burden differed from place to place.

Villages within one district differed as well. Among the 17 villages of Kepsut, counting direct tax only, the lowest, Sebiller, stood at 12.8 per cent and the highest, Umurlar, at 24.4 per cent, close to twice as much.

Russia: little went to the state, much went to the lord

For Moscow province in 1811, according to a province-wide income table reconstructed by two researchers, the direct taxes that state peasants paid to the state came to about a quarter of their pre-tax income, while serfs paid the state only about 7 per cent. Serfs also owed their lords quitrent (obrok) or labour service, which is not included in the 7 per cent.

Within the state peasants' quarter is an item called the "state peasants' quitrent tax" (obrochnaya podat'). It is the equivalent of the quitrent that serfs paid their lords, except that state peasants paid it to the state. Read this way, the main difference between the two kinds of peasant lies in who received this share, not in one being lighter and the other heavier. This is my reading; the authors do not say so.

Japan: one village in two years

Tawara village in Yoshino district, Yamato province, left two village-wide household-by-household reports of income and expenditure, one written in 1808 and one in 1837; both were written by the village to ask for a reduction of the annual tax. In the original, the annual tax and the village's own expenses ("jōnō and ko-nyūyō") appear as a single total and cannot be separated. That total as a share of the village's income was 39 per cent in 1807 and 50 per cent in 1836, the year of the Tenpō famine.

This is much higher than economic historians' estimate for Japan as a whole at the end of the Edo period, which on a collected-revenue basis was 16 to 17 per cent of agricultural output. The reasons cannot be separated at present: the numerator includes village expenses; the reports were written to ask for relief, so income may have been under-reported; and which lord held the village has not yet been established.

England and Qing China: filled in for now from secondary studies

In these two places the original registers do not give the output of the same place, so for now we use figures from secondary studies that name a place and a date.

England: in 364 villages of four Midland counties covered by the Hundred Rolls of 1279–80, the recorded rent of villein tenants comes to about 10.5 per cent of the gross output of an 18-acre holding. The tithe for the whole kingdom in 1340–41 was roughly one twenty-fifth of total rural income. Around 1300 the land income of all lords together amounted to 18 to 23 per cent of rural income, but about three-tenths of that was the profit of the lords' own demesne farming, not payments by tenants, so it is shown separately as a combined total. The king's lay subsidies were assessed on movable goods, and there is no figure for them as a share of output.

Qing China, around Suzhou: in about 1885–1911 the land tax actually collected around Suzhou and Shanghai took 8 to 10 per cent of the value of land output, higher than the 2 to 4 per cent of most provinces. For the landlord's share: after the rent reduction of the Tongzhi reign (1865), the nominal rent per mu in Suzhou rent-bursary ledgers was mostly between 1.0 and 1.25 shi of rice, and about seven-tenths of the nominal rent was actually collected. A contemporary, Tao Xu, wrote that a mu yielded at most three shi of rice. Taking three shi as the upper limit of the harvest, the rent actually collected was at least 23 to 29 per cent of the harvest.

What these figures can and cannot show

They show that, judged by the treasury, all these states took little, but judged by what the original registers record under the peasants' names, the burden was not light: 20 to 30 per cent in most Ottoman places, about a quarter for Russian state peasants, and 40 to 50 per cent in Tawara village in Japan. And within a single state, one place could differ from another several times over.

They have four limits. First, the denominators are all registered income, not measured output. Ottoman incomes were mostly worked back from the tithe; some include the tithe and some are net of the tithe or of the sharecropping split. These figures can be set side by side but cannot be averaged into one number. Second, the direct tax in the Ottoman registers is the previous year's payment, while income is the year of registration, a year apart. Third, the English and Qing figures come from secondary studies, not original registers, and the Suzhou landlord's share is only a lower bound. Fourth, places differed so widely that no figure from one place can be extended to another.

Method and sources

Each figure in the table was either computed by us, by adding up the transcribed original registers household by household and dividing, or computed by the original researcher and checked by us against the printed page; a few rows are figures quoted at second hand in secondary studies, and are marked as such. The full 63-row table gives, row by row, the definition of numerator and denominator, the source and page, and how far the figure was checked, and can be downloaded.

Main sources: Demir 2016 (Kepsut, 17 villages); Kolay 2014 (Bezirgân); Atar 2026 (Tivnik); Küçükkalay and Efe 2006 (Alpu); Seyyar 2019 (Otmanlı); Pehlivan 2023 (Ağsaklar); Sağlam 2022 (Saruhanlar); Hançer 2026 (Rumnos); Güran 1985 (Ödemiş); Zabun 2021 (Kilmigad); Üçtepe 2006 (Artukabad); Koyuncu and Küçükkalay 2016; Yükçü et al. 2014 (Tire); İlter 2022 (Buldan); Yereli et al. 2015 (Sorgun); Gölçek and Köktaş 2019 (Niğde); Efe 2016 (sub-provinces, 1847); Korchmina and Malinowski 2026 (Moscow province); Morozov 2010 (Rostov); Kinoshita Mitsuo 2013, 2018 (Tawara); Campbell 2005, Kanzaka 2002 (England); Wang Yeh-chien 1973, Natsui Haruki 1989, 1998 (Suzhou); Karaman and Pamuk 2010, Pamuk 2009 (Ottoman treasury).

Data table

All the data in the chart, 63 rows, giving for each row the definition of numerator and denominator, who computed it, the source and how far it was checked.

Download CSV (English)