Data on landholdings of rural households are available from NSS surveys on Land and Livestock Holdings and from Agricultural Censuses. Limited data on ownership and operational holding of land are also available on other NSS surveys (for example, NSS Surveys on Consumer Expenditure and NSS Surveys on Employment and Unemployment).

NSS Surveys on Landholdings

Rawal(2013)\cite{nssland2013} on data from NSS Surveys on Land and Livestock holdings:

Surveys of Land and Livestock Holdings are conducted by NSSO once in ten years. In these surveys, detailed information is collected on various aspects of land and livestock holdings. These include collection of plot-wise data on land use and on tenurial status. There are two limitations of Land and Livestock Surveys. First, these surveys are conducted only once in ten years. The last survey was conducted in 2002-03 and nothing more recent is available. Secondly, these surveys collect data on all types of land, including homestead lands, and published NSSO reports do not distinguish between these. \cite{meadcain1983} and \cite{epwland2008} Have pointed out that this creates enormous confusion in estimation of extent of landlessness and inequality in land distribution. \cite{meadcain1983} pointed out that, “change and ambiguity in important concepts and definitions and inconsistency and incompleteness in published tabulations prevent confident estimation of the percentage of landless households at different points in time. Far from being able to discern the rate of change in landlessness, we find that the direction of change remains uncertain.” Availability of household-level data from the 48th round survey made it possible to separate different types of land. However, only two rounds of surveys have been conducted since then. Also, \cite{epwland2008} has pointed out that the data from these two rounds are not strictly comparable because of a change in the way data on cultivation on homestead lands were recorded.

Reading Unit-level Data

48th round

An awk script to read NSS 48th round unit-level data is here. This script will generate csv files that can be imported into any statistical software capable of handling large datasets.

59th round

An awk script to read NSS 48th round unit-level data is here. This script will generate csv files that can be imported into any statistical software capable of handling large datasets.

Data from NSSO Surveys on Employment and Unemployment/Consumer Expenditure

Extracts from Rawal(2013) \cite{nssland2013} on these data:

There are three items of data in Schedule 10 (on Employment and Unemployment) and Schedule 01 (on Consumer Expenditure) surveys pertaining to landholdings. These pertain to:

  • Land owned by the household,
  • Land possessed by the household, and
  • Land cultivated by households.

Conceptually, data on land owned correspond to ownership holding of land and data on land possessed correspond to operational holding of land. However, in the NSSO surveys, these items deal with not just agricultural land but with all types of land, including homestead land. Aggregation of physical extent of plots of land that are very dissimilar in terms of their use, their productive potential and their market value makes these variables unusable.

On the other hand, data on land cultivated pertain to agricultural land only, and corresponds to net sown area. Strictly speaking, land cultivated is different from operational holding of land as it does not include land that is left fallow.

There are three important limitations of this dataset.

First, like all sources of data on landholdings, there is a considerable under-reporting of land by households, in particular, by households that own large amounts of land. An implication of under-reporting of land, particularly by large landowners and by lessors of land, is that the estimates of total land cultivated from the NSSO surveys of employment and unemployment account for only about 70 per cent of net sown area at the national level.

In the context of under-reporting of land, it may also be noted that, since the extent of under-reporting is supposed to be large for large landholdings, these data are likely to under-estimate inequality in ownership and operational holding of land.

Secondly, like all official sources on landholdings in India, the NSSO surveys on employment and unemployment also capture very little incidence of tenancy. Given that most States of India have legislation under which leasing of agricultural land is prohibited or regulated, most tenancy contracts in rural India are informal and oral. Incidence of such tenancy is concealed and does not appear either in official records or in official surveys. This problem affects NSSO surveys just as it affects all other sources of data on land.

An inconspicuous but important change has been introduced since 2004-05 in questions pertaining to land in Schedule 1.0 and Schedule 10 of NSSO surveys. Until the 1999-2000 survey, the NSSO schedules were designed to record the information on extent of land up to two decimal places. As a result, the smallest land holding recorded was 0.01 hectares. Any household having land less than 0.01 hectares was treated as landless. In all tabulations in NSSO reports also, the smallest land size was 0.01 hectares and households having less than 0.01 hectares were treated as landless. In the 61st round (for 2004-05), a small but important change was introduced in the NSSO schedules. From the 61st round onwards, the Schedule 1.0 and Schedule 10.0 started recording data pertaining to land owned, land possessed and land cultivated up to three decimal places. As a result, the smallest size of holding recorded became 0.001 hectares. Correspondingly, from 2004-05, NSSO reports started to classify households having land between 0.001 hectares and 0.01 hectares also as owning/possessing/cultivating land, and only households having less than 0.001 hectares of land as landless households.

This change resulted in a huge drop in proportion of landless households in the estimates based on these surveys. The effect was particularly significant in estimates of landlessness in terms of ownership or possession of land because these variables included homestead land. As a result of this change, households having only tiny amounts of homestead land got classified as landowning households. This change resulted in lowering the proportion of households that did not own any land by about 20 percentage points and the proportion of households that did not possess land by about 23 percentage points in each of the rounds since 2004-05.

Unless accounted for, this change makes data from surveys prior to 61st round incomparable with data from 61st round survey and thereafter.

Agricultural Censuses

Data from Agricultural Censuses pertain only to operational holdings; there are no data on ownership of land.

Recent data from Agricultural Censuses and documents about these data are available at this website.

It may be mentioned that Agricultural Census data on landholdings are extremely problematic. Some of the main problems are:

  • Agricultural Censuses are conducted as part of the World Censuses of Agriculture coordinated by the Food and Agricultural Organization of the United Nations. However, unlike the World Agricultural Censuses, Indian Agricultural Census “does not include those holdings which are not operating any agricultural land and are engaged exclusively in livestock, poultry and fishing etc.(Source) As a result, Agricultural Censuses in India do not give an data on landlessness even in terms of access to operational holdings.
  • In a majority of States, agricultural censuses are based on retabulation of land records. Land records are not only plagued by benami registrations but are also often not updated when ownership of land is transferred because of sale, inheritance or partitioning.\cite{nssland2013}
  • Land records are usually individual-based and do not identify a household. In the survey manuals for Agricultural Censuses, individual holding is defined at the level of the household: “An individual holding may be operated by one person alone or jointly by a group of persons but the basic condition is that these persons must belong to the same household, i.e., they must be residing together and taking their food from a common kitchen. (Source)” However, the schedule for data collection for Agricultural Censues is designed to record a single individual as an “operational holder”. There is no clearly stated method that is adopted to transform this individual owner/cultivator-level information into household-level information, if this is done at all. The village-level field officer is asked the verify details about actual de-facto status on the basis of “personal knowledge”. The instructions in this respect are vague, non-objective and are unlikely to have been implemented with any seriousness. It may also be noted that over the last few decades, village-level field workers are transferred from one village to another, and may not even have much “personal knowledge” about who actually operates each plot of land in the village (which could easily run into several thousands of plots in any Indian village).
comments powered by Disqus