Agricultural Statistics & Data Science · Sample surveys and official agricultural statistics · also Official Agricultural Statistics, Indian Official Statistics, Agricultural Statistics System
Who collects which number and how: land-use and crop statistics, the agricultural census and input survey, livestock census and integrated sample survey, fishery and horticulture statistics, cost of cultivation, price and market statistics, and the international frameworks (FAO World Programme for the Census of Agriculture, Global Strategy, FAOSTAT, SDG indicators).
In the exams: AICE-JRF Ag Stat Unit 5 'agricultural statistics system: land use, crop estimation, livestock and fishery statistics' (heavy); UPSC ISS Statistics IV official statistics (heavy); SSC CGL JSO (present); state statistical-officer exams (heavy)
Agricultural statistics are the numbers on which food policy, prices, trade, insurance and research depend: how much land is cultivated, what crops are grown, how much is produced, how many animals there are, what farming costs and what farmers earn. This subject describes the systems that produce those numbers, first in general terms and through the international frameworks of FAO, the Global Strategy and the SDGs, and then in detail for India's system, with comparisons to the United States, the European Union and other countries.
It covers national statistical systems and India's MoSPI/NSO and Directorate of Economics and Statistics; land-use statistics and the nine-fold classification; the Timely Reporting Scheme and advance estimates; the agricultural census and input survey; the livestock census and integrated sample survey; fishery and horticulture statistics; the cost of cultivation scheme; prices, wages and market arrivals; the NSS situation assessment surveys; the FAO World Programme for the Census of Agriculture and FAOSTAT; the Global Strategy, 50x2030, SDG indicators and quality frameworks; and other national systems.
The statistical methods behind these systems are covered elsewhere: crop-cutting experiments and yield estimation in crop-estimation-surveys, questionnaire and field operations in survey-design-operations, sampling in the sampling subjects, and index numbers and national accounts in economic-statistics-index-numbers. Institutional details and release schedules change; figures quoted here are rounded and given as illustrations, and current publications should be consulted for official values.
Every country's agricultural numbers come from a system, not from a single survey. A national statistical system is the set of organisations that collect, compile and publish official statistics, together with the laws, standards and coordination arrangements that bind them. Systems are centralised (one national statistical office produces most statistics, as in Canada, Australia or the Netherlands) or decentralised (several ministries produce statistics in their own fields under coordination, as in the United States and India). Agricultural statistics are often produced by the ministry of agriculture rather than the national statistical office, because they depend on administrative channels such as land records, extension staff and market committees. Whatever the arrangement, the UN Fundamental Principles of Official Statistics (adopted by the UN Statistical Commission in 1994 and endorsed by the UN General Assembly in 2014) set the norms: relevance, impartiality, professional independence, confidentiality, transparency of methods, and use of international standards.
India's statistical system is decentralised. The Ministry of Statistics and Programme Implementation (MoSPI) is the nodal agency; its National Statistical Office (NSO) was formed in 2019 by merging the Central Statistics Office and the National Sample Survey Office. The NSO compiles national accounts and price indices, conducts the National Sample Survey through its survey design and field operations wings, coordinates standards, and supervises the quality of crop statistics through the Improvement of Crop Statistics scheme. The National Statistical Commission, set up in 2005 on the recommendation of the Rangarajan Commission (2001), advises on priorities and standards. The Collection of Statistics Act, 2008 provides legal backing for surveys and for confidentiality, as discussed in survey-design-operations.
Most primary agricultural statistics are compiled by the Directorate of Economics and Statistics (DES) of the Ministry of Agriculture and Farmers Welfare. Its responsibilities have included land-use statistics, area and production estimates of principal crops (including the advance estimates), the cost of cultivation scheme, farm harvest prices, wholesale prices and wages, the index numbers of agricultural production and of terms of trade, and the annual publication Agricultural Statistics at a Glance. The agricultural census is conducted by the ministry's agriculture census division. Livestock and fisheries statistics belong to the Department of Animal Husbandry and Dairying and the Department of Fisheries, and horticulture statistics are compiled through the National Horticulture Board. At state level, the State Agricultural Statistics Authority (often the Director of Land Records, Agriculture or Economics and Statistics) and the revenue machinery (the village accountant, patwari or karnam, who records crops in each field during the girdawari or crop inspection) supply the basic data.
A decentralised system has to manage overlaps and gaps: crop area from land records and from remote sensing, production from crop cutting and from trade sources, livestock products from the integrated sample survey and from industry. Consistency checks between sources, common concepts and a published calendar of releases are the main tools. Reviews such as the report of the Committee on Agricultural Statistics and the National Statistical Commission's own reports have repeatedly identified delays in land records, weak crop-cutting supervision and limited use of new technology as problems, which explains the recent push for digital crop surveys and remote sensing discussed in crop-estimation-surveys.
Land-use statistics in India are compiled from village land records by the revenue machinery, aggregated to districts and states and published by DES. The reporting area is the area for which land-use data are available; it is smaller than the geographical area surveyed by the Survey of India, mainly because some hilly, forest and border areas lack village papers. For areas not covered by cadastral survey, estimates are based on other sources. Users should therefore always state whether a percentage refers to geographical or reporting area.
The present classification was recommended by the Technical Committee on Coordination of Agricultural Statistics in 1949 and adopted from 1950-51, replacing the older five-fold classification. The nine classes are: (1) forests; (2) area under non-agricultural uses (buildings, roads, railways, water bodies); (3) barren and unculturable land (mountains, deserts); (4) permanent pastures and other grazing land; (5) land under miscellaneous tree crops and groves not included in net area sown; (6) culturable waste land (available for cultivation but not cultivated for more than five years); (7) fallow land other than current fallows (left uncultivated for one to five years); (8) current fallows (left uncultivated during the current year only); and (9) net area sown. Classes 2 and 3 together form 'land not available for cultivation'; classes 4, 5 and 6 are 'other uncultivated land excluding fallow land'; classes 7 and 8 are 'fallow lands'.
Net area sown counts each piece of land once in an agricultural year (July to June in India), however many crops it carries. Area sown more than once is the land cropped in more than one season; gross cropped area (total cropped area) counts land once for each crop grown. Cropping intensity is gross cropped area divided by net area sown times 100: if net area sown is 140 million ha and gross cropped area 196 million ha, cropping intensity is 140%. India's net area sown has been roughly 140 million ha for decades, so growth in output has come from intensity and yields. Land-use statistics also include irrigated area by source (canals, tanks, tube wells, other wells, other sources), net and gross irrigated area, and area under each crop.
FAO's land-use statistics (in FAOSTAT) use a different classification based on the World Programme for the Census of Agriculture: agricultural land (cropland, comprising arable land and permanent crops, plus permanent meadows and pastures), forest land and other land. Converting India's classes to FAO categories requires judgement; for example, culturable waste and fallow land other than current fallows may or may not be counted as arable land. Remote-sensing land-use and land-cover products, such as those of the National Remote Sensing Centre, provide an independent picture but follow land-cover rather than land-use logic.
Suppose a district has a reporting area of 500,000 ha: forests 60,000; non-agricultural uses 40,000; barren and unculturable land 20,000; permanent pastures 15,000; miscellaneous tree crops 5,000; culturable waste 10,000; other fallows 15,000; current fallows 35,000; and net area sown 300,000. Land not available for cultivation is 60,000 ha, fallow land is 50,000 ha, and the nine classes add up to the reporting area. If 120,000 ha are sown more than once, gross cropped area is 420,000 ha and cropping intensity is 140%. A rise in current fallows in a drought year, with a corresponding fall in net area sown, is a typical pattern that land-use statistics reveal.
In most states, called temporarily settled or land-record states, the village accountant inspects every field each season (girdawari) and records the crop in the village register (khasra). Complete enumeration is slow, so the Timely Reporting Scheme (TRS), started in the late 1960s, requires a 20% random sample of villages to be inspected on priority and reported quickly; state area estimates are built from this sample. In states without such village records (historically Kerala, Odisha and West Bengal, the permanently settled states), the scheme for Establishment of an Agency for Reporting Agricultural Statistics (EARAS) created a staff to conduct a sample survey of crop area in a 20% sample of villages. In the north-eastern hill states, estimates historically relied on conventional methods and eye estimation, with gradual extension of sampling.
Production is estimated as area times average yield, with yield from crop-cutting experiments under the General Crop Estimation Surveys planned by NSO and ICAR-IASRI, as treated in crop-estimation-surveys. The Improvement of Crop Statistics (ICS) scheme has the NSO's field staff and state supervisors check a sub-sample of the TRS villages and crop-cutting experiments, reporting discrepancies in area records and in experiment procedure. ICS reports have long found substantial discrepancies in the recording of crops by village accountants, which is one reason for the shift to digital, geo-referenced crop surveys.
Because policy needs production figures before final estimates are possible, DES releases a sequence of advance estimates for the agricultural year, beginning with a first estimate of kharif crops early in the season and ending with final estimates after crop-cutting results are complete; historically there were four advance estimates followed by final estimates, and the release schedule has been revised in recent years. Early estimates rely on state reports, sown area returns, weather and crop condition, and remote sensing; later ones increasingly on TRS area and crop-cutting yields. Revisions between advance and final estimates are an important quality indicator, and analysts should always note which estimate they are using.
The FASAL programme (Forecasting Agricultural output using Space, Agro-meteorology and Land-based observations) and the Mahalanobis National Crop Forecast Centre (MNCFC) produce remote-sensing-based pre-harvest forecasts of the area and production of major crops, which complement the official advance estimates. Similar programmes exist elsewhere: the European Commission's MARS crop yield forecasting system, USDA's satellite and survey-based forecasts, and the GEOGLAM crop monitor for global markets.
Area estimates depend on the village accountant actually visiting fields at the right time, recognising crops correctly, and recording mixed crops according to the prescribed rules for apportioning area among component crops. When visits are skipped, entries are copied from the previous season, which perpetuates errors. Sample checks under ICS, comparisons with remote-sensing estimates, and geo-tagged photographs in digital crop surveys are the main safeguards. Users comparing official area with satellite estimates should remember that the two may also differ in definitions, for example in the treatment of small plots, field bunds and intercrops.
A census of agriculture is a statistical operation for collecting data on the structure of agriculture (the number and size of holdings, land tenure, land use, crop area, irrigation, livestock and labour) covering the whole or a significant part of a country. Its unit is the agricultural holding, defined by FAO as an economic unit of agricultural production under single management comprising all livestock kept and all land used wholly or partly for agricultural production, without regard to title, legal form or size. India uses the operational holding: all land used wholly or partly for agricultural production and operated as one technical unit by one person alone or with others, without regard to title, legal form, size or location. Operational holdings differ from ownership holdings because land is leased in and out.
India has conducted an agricultural census every five years since 1970-71, in line with the FAO World Programme. Data for land-record states are retabulated from village records for all holdings, while in non-land-record states a sample survey of villages is used. The 2015-16 census counted about 146 million operational holdings with an average size of about 1.08 ha; marginal (below 1 ha) and small (1 to 2 ha) holdings together made up about 86% of holdings. The standard size classes are marginal (below 1 ha), small (1 to 2 ha), semi-medium (2 to 4 ha), medium (4 to 10 ha) and large (10 ha and above). The eleventh census, with reference year 2021-22, introduced digital data collection using tablets and smartphones. Results are tabulated by size class, social group and gender of the operational holder.
The input survey follows each agricultural census in a sample of villages and holdings. It collects data on the use of inputs by size class of holding: fertilisers and manures, certified and high-yielding seeds, pesticides, agricultural machinery and implements, livestock, and agricultural credit. Because it uses the census holdings as its frame, it provides input use by holding size that no other source gives, which is valuable for policy on subsidies and credit.
The census series reveals long-run structural change: in India the number of holdings has increased and average size has fallen census after census, with the share of marginal holdings rising. Land Gini coefficients (treated in economic-statistics-index-numbers) can be computed from the size distribution. Comparisons across censuses require attention to changes in concepts, coverage and methods, for example the move from complete retabulation to sampling in some states.
Agricultural census results are used to plan credit and input supply by size class, to target schemes for small and marginal farmers, to measure the participation of women as operational holders, and as a frame for later surveys such as the input survey and cost studies. Their limitations include the long interval between censuses, the time taken to publish results, the dependence on land records that may not reflect actual operation (unrecorded tenancy is common), and the omission of landless livestock keepers unless special provisions are made.
India's livestock census has been conducted roughly every five years since 1919-20 by the Department of Animal Husbandry and Dairying with state animal husbandry departments, by complete enumeration of households, household enterprises and non-household enterprises such as dairy farms, poultry farms and gaushalas. It records livestock (cattle, buffalo, sheep, goats, pigs, horses and ponies, mules, donkeys, camels, mithun and yak) and poultry by species, breed, age, sex and use, together with equipment such as chaff cutters. The 20th census (2019) reported total livestock of about 536 million and poultry of about 852 million; it was the first to collect data on tablets and to record breed-wise data for many species. The 21st census began in 2024. Results feed the national accounts, disease-control planning and breeding policy.
Production of major livestock products (milk, eggs, meat and wool) is estimated by the Integrated Sample Survey (ISS), conducted by state animal husbandry departments under a design developed with ICAR-IASRI. The year is divided into three seasons (summer, rainy and winter), and in each season a sample of villages is selected within strata; in each village, households are listed with their animals and a sample of households is visited to record actual milk yields, eggs laid, animals slaughtered and wool clipped. Milk production is estimated as the number of animals in milk times the average yield per animal in milk per day times the number of days. For example, 2.0 million cows in milk averaging 8 kg per day over 365 days produce 2.0 x 8 x 365 = 5,840 million kg, or about 5.84 million tonnes. Results are published in the annual Basic Animal Husbandry Statistics. India has been the world's largest milk producer for many years, with production exceeding 230 million tonnes in 2022-23 by these estimates.
Livestock statistics face specific problems: animals move, ages are hard to determine, household slaughter and unregistered slaughter are under-recorded, and the census is used as the frame for the sample survey long after it was conducted. Meat estimates in particular rely on registered slaughterhouses plus survey estimates of unregistered slaughter and are generally regarded as less reliable than milk estimates. Internationally, FAO's livestock statistics and the livestock modules of the World Programme for the Census of Agriculture provide standards.
The livestock census provides the numbers of animals by category, which serve as the frame and the multipliers for the integrated sample survey; the survey provides yields per animal. Between censuses, animal numbers are projected using growth rates, so errors in projected populations pass directly into production estimates. When a new census differs sharply from the projection, production series may need revision, and users of long series should look for such breaks.
Marine fish landings in India are estimated by ICAR-Central Marine Fisheries Research Institute (CMFRI) using a stratified multistage random sampling design developed from the 1950s. Each maritime state's coastline is divided into zones; landing centres are stratified by size; landing-centre days are selected within each month; and the catch of a sample of craft is observed and species composition recorded during observation periods. Total landings are estimated by raising the sample to all fishing days and centres. The census of marine fishing villages and craft provides the frame. The Department of Fisheries compiles national fish production including state estimates and publishes the Handbook on Fisheries Statistics.
Inland capture and culture fisheries are spread across rivers, reservoirs, tanks, ponds, floodplain wetlands and brackish waters, and are harder to measure. States use sample surveys, departmental records and auctions; quality varies widely. Methodologies for reservoir and pond culture estimation developed with ICAR-IASRI and ICAR-CIFRI have been recommended. India's total fish production has risen strongly, with inland aquaculture now accounting for most of it. FAO's FishStat and the State of World Fisheries and Aquaculture (SOFIA) report give global statistics on capture and aquaculture.
Horticulture statistics (fruits, vegetables, flowers, plantation crops, spices, aromatic and medicinal plants) have historically been weaker than those for field crops because these crops are grown in small, scattered plots, often mixed, harvested repeatedly and not covered by crop-cutting experiments. The National Horticulture Board compiles area and production data from state horticulture departments. The CHAMAN programme (Coordinated Horticulture Assessment and Management using geoiNformatics), implemented by MNCFC with ICAR-IASRI and others, used remote sensing and sample surveys to estimate area and production of major horticultural crops in selected districts. In recent years India's estimated horticultural production has exceeded foodgrain production, which makes the quality of these statistics important for policy.
The Comprehensive Scheme for the Study of Cost of Cultivation of Principal Crops in India (CCS) has been run by DES since 1970-71 through state agricultural universities and other agencies. It collects farm-level data on inputs, outputs and their prices for principal crops, and the results are the main input to the Commission for Agricultural Costs and Prices (CACP) in recommending minimum support prices (MSP). The design is a three-stage stratified random sample: tehsils (or clusters of villages) are first-stage units stratified by agro-climatic zone, a village or cluster of villages is selected in each, and about ten operational holdings are selected per cluster, two from each of five size classes. Data are recorded by cost accounting throughout the year by resident field staff.
CCS reports cost under nested concepts. Cost A1 includes all actual expenses in cash and kind incurred by the owner-operator: hired human labour, bullock and machine labour (hired and owned), seed, fertilisers, manures, pesticides, irrigation charges, land revenue, depreciation of implements and farm buildings, interest on working capital and miscellaneous expenses. Cost A2 = A1 + rent paid for leased-in land. Cost B1 = A1 + interest on the value of owned fixed capital assets (excluding land). Cost B2 = B1 + rental value of owned land (net of land revenue) + rent paid for leased-in land. Cost C1 = B1 + imputed value of family labour. Cost C2 = B2 + imputed value of family labour. Cost C2* values family labour at the statutory minimum wage or the actual market wage, whichever is higher, and Cost C3 = C2* + 10% of C2* as the value of management. Cost A2 + FL (A2 plus imputed family labour) is also reported.
Since 2018-19 the government has announced MSPs for kharif and rabi crops at no less than 1.5 times Cost A2 + FL; for example, if A2 + FL for a crop is Rs 1,400 per quintal, the MSP is at least Rs 2,100 per quintal. Farmer organisations have argued that the benchmark should be C2, following the National Commission on Farmers' recommendation of MSP at least 50% above the weighted average cost of production. Statistical issues include the age of the sample design, the imputation of family labour and owned land, the lag between data collection and use, and the representativeness of a small number of holdings per crop and state. Similar farm-cost surveys elsewhere include the USDA's Agricultural Resource Management Survey and the EU Farm Accountancy Data Network.
Cost per hectare is converted to cost per quintal by dividing by yield, after deducting the value of by-products such as straw, which are credited against cost. A crop with C2 cost of Rs 60,000 per hectare, by-product value of Rs 4,000 per hectare and a yield of 40 quintals per hectare has a cost of production of (60,000 - 4,000)/40 = Rs 1,400 per quintal. Because yields vary greatly between years and holdings, cost per quintal is more variable than cost per hectare, and averages are weighted by area or production according to their purpose.
Agricultural price statistics serve different purposes and must not be confused. Farm harvest prices, compiled by DES from state reports, are the average wholesale prices at which a commodity is disposed of by the producer to the trader at the village site during the harvest period; they are used to value crop output in the national accounts. Wholesale prices are collected weekly from selected markets for the wholesale price index and for market intelligence. Retail prices are collected for the consumer price indices. Minimum support prices are policy prices, not market observations. Producer prices in FAOSTAT are, in principle, farm-gate prices.
Regulated markets under the state Agricultural Produce Market Committee (APMC) Acts record arrivals and prices of commodities. The Directorate of Marketing and Inspection's Agmarknet portal publishes daily arrivals and minimum, maximum and modal prices from a large number of markets, and the e-NAM electronic platform adds trade data. Arrivals are not production: they include produce from other districts, exclude sales outside regulated markets and on-farm consumption, and can be double-counted when produce moves between markets. Arrival and price data are nonetheless invaluable for market monitoring, price forecasting and the analysis of price spreads.
Agricultural wage statistics include DES's Agricultural Wages in India, compiled from reports from selected centres for operations such as ploughing, sowing, weeding, harvesting and other agricultural labour, separately for men, women and children, and the Labour Bureau's Wage Rates in Rural India, collected monthly from sample villages for agricultural and non-agricultural occupations. Labour Bureau also compiles the consumer price indices for agricultural and rural labourers used to compute real wages, as described in economic-statistics-index-numbers.
The FAO Food Price Index tracks international prices of a basket of food commodities. FAO's GIEWS tool monitors domestic food prices in many countries, and the indicator of food price anomalies (SDG 2.c.1) identifies abnormally high prices. USDA NASS publishes monthly Agricultural Prices, including the prices-received and prices-paid indices used in the US parity concept.
The NSS has conducted Situation Assessment Surveys (SAS) of farmers or agricultural households in its 59th round (2003), 70th round (2013) and 77th round (2019). They complement the agricultural census by giving household-level information on income, consumption, indebtedness, access to technology and services, and awareness of schemes. The 77th round, conducted from January to December 2019 with the agricultural year July 2018 to June 2019 as reference, was combined with the Land and Livestock Holdings survey and collected data in two visits so that both kharif and rabi seasons were covered.
In the 77th round an agricultural household was defined as one receiving more than Rs 4,000 as value of produce from agricultural activities (cultivation of field crops, horticulture, fodder, plantation, animal husbandry, poultry, fishery, piggery, bee-keeping, vermiculture, sericulture and so on) during the last 365 days, with at least one member self-employed in agriculture as principal or subsidiary status. The previous rounds used different thresholds, so comparisons across rounds must adjust. Income from crop production is computed from the value of output minus paid-out expenses, so imputed family labour and owned land are not deducted; it is therefore closer to a farm business income than to a profit.
The 77th round estimated about 93 million agricultural households and an average monthly income per agricultural household of a little over Rs 10,000 in 2018-19, with wages and livestock forming large shares alongside crop production. It also measured the share of households indebted and the sources of loans. These figures were central to the debate on the goal of doubling farmers' incomes. Users must remember the sampling error at state level, the different definitions in each round, and the fact that income questions are subject to large response errors, as discussed in survey-design-operations.
The World Programme for the Census of Agriculture (WCA), coordinated by FAO, recommends that each country conduct an agricultural census at least once in each ten-year round and provides concepts, definitions, core items and tabulations so that results can be compared. It began with the 1930 round organised by the International Institute of Agriculture, FAO's predecessor. The WCA 2020 round covers censuses conducted between 2016 and 2025; FAO has prepared guidance for the 2030 round. The programme recognises several census modalities: the classical approach (a single complete or large-sample operation), the modular approach (a core module by complete enumeration with supplementary sample-based modules), the integrated census and survey approach (the census as a base for a programme of annual and periodic surveys), and the use of registers as a source of census data.
WCA guidance lists essential items such as the identification and location of the holding, the legal status of the holder, the sex and age of the holder, the area of the holding by land use and tenure, irrigation, crops, livestock numbers and the use of inputs, grouped into themes such as land, irrigation, crops, livestock, agricultural practices, services, demographic and social characteristics, work, household food security, aquaculture, forestry and environment. The 2020 round placed more emphasis on gender, on the holder and on links with population censuses and other surveys.
FAOSTAT, FAO's statistical database, provides time series from 1961 for more than 245 countries and territories on production, trade, prices, land use, inputs, emissions, food security indicators and investment. Data come from country questionnaires, supplemented by imputation where countries do not report, which users should check through the flags attached to each value. Food balance sheets present a country's supply and use of each commodity for a year: domestic supply = production + imports - exports - increase in stocks (or + decrease), and supply is distributed among feed, seed, processing, losses, other uses and food. For example, production 100, imports 20, exports 10 and a stock increase of 5 give a domestic supply of 105. Dividing food use by population and converting to calories, protein and fat gives per capita food supply, an input to the prevalence of undernourishment.
Food balance sheets measure food available for consumption at the national level, not what people actually eat, and they say nothing about distribution between households. Their quality depends on production, trade and stock statistics, and stocks and losses are often poorly measured. Household consumption surveys measure food acquisition directly but miss food eaten outside the home and have their own errors. FAO combines information from both kinds of source in estimating undernourishment.
The Global Strategy to Improve Agricultural and Rural Statistics was endorsed by the UN Statistical Commission in 2010, prepared by the World Bank, FAO and the UN Statistics Division in response to the decline of agricultural statistics in many developing countries. It rests on three pillars: a minimum set of core data that all countries should provide; the integration of agriculture into national statistical systems, through a master sampling frame, an integrated survey framework and data management systems; and the sustainability of agricultural statistics through governance and capacity building, with countries preparing Strategic Plans for Agricultural and Rural Statistics (SPARS). FAO hosts the Global Office, and research outputs include guidelines on master sampling frames, crop yield measurement, cost of production statistics, post-harvest losses and the use of remote sensing.
The 50x2030 Initiative to Close the Agricultural Data Gap, launched in 2019 and implemented by FAO, IFAD and the World Bank, aims to help 50 low- and lower-middle-income countries build integrated agricultural survey systems by 2030. It builds on FAO's AGRISurvey programme and the World Bank's Living Standards Measurement Study - Integrated Surveys on Agriculture (LSMS-ISA), combining an annual core agricultural survey with rotating modules on production methods, economics, labour and the environment, and linking agricultural and household data. Its emphasis is on data use as well as data production.
The Sustainable Development Goals include many indicators that rely on agricultural statistics, with FAO as custodian agency for about two dozen. Examples are 2.1.1 prevalence of undernourishment; 2.1.2 prevalence of moderate or severe food insecurity measured by the Food Insecurity Experience Scale; 2.3.1 volume of production per labour unit by classes of enterprise size; 2.3.2 average income of small-scale food producers; 2.4.1 proportion of agricultural area under productive and sustainable agriculture; 2.5.1 and 2.5.2 on genetic resources; 2.a.1 the agriculture orientation index for government expenditure; 2.c.1 food price anomalies; 6.4.1 water-use efficiency and 6.4.2 level of water stress; 12.3.1 food loss and waste indices; and 15.1.1 forest area as a proportion of land area. The agriculture orientation index is the agriculture share of government expenditure divided by the agriculture share of GDP: a share of 6% of expenditure and 18% of GDP gives 0.33, indicating less orientation to agriculture than its economic weight. India's NSO publishes a National Indicator Framework that adapts the global indicators.
Quality frameworks translate principles into checkable criteria. Examples are the UN National Quality Assurance Framework (NQAF) manual, the IMF Data Quality Assessment Framework (DQAF) and its dissemination standards (India subscribes to the Special Data Dissemination Standard), the European Statistics Code of Practice, and FAO's Statistics Quality Assurance Framework. Common dimensions are relevance, accuracy and reliability, timeliness and punctuality, accessibility and clarity, coherence and comparability, and the institutional environment (independence, mandate, resources and confidentiality). Applying them to agricultural statistics means documenting sources and methods, publishing revision analyses of advance estimates, and reporting sampling and non-sampling errors.
In the United States the National Agricultural Statistics Service (NASS) of the Department of Agriculture produces most agricultural statistics, in cooperation with state departments of agriculture. Since 1997 NASS has conducted the Census of Agriculture, previously run by the Census Bureau, every five years for years ending in 2 and 7; a farm is defined as any place from which $1,000 or more of agricultural products were produced and sold, or normally would have been sold, in the census year. Its survey programme rests on a combination of an area frame (used in the June Agricultural Survey, which also measures list-frame incompleteness) and a list frame of farmers, a classic multiple-frame design. Major reports include Crop Production, Acreage, Grain Stocks, Hogs and Pigs, Agricultural Prices and the weekly Crop Progress. Market-sensitive reports are prepared under strict security ('lockup') and released at announced times to prevent advance knowledge. NASS also produces the Cropland Data Layer from satellite imagery and, with the Economic Research Service, the Agricultural Resource Management Survey of farm finances and practices.
In the European Union, national statistical institutes collect agricultural statistics under EU regulations and Eurostat harmonises and publishes them. The Integrated Farm Statistics regulation (Regulation (EU) 2018/1091) governs the farm structure survey and the 2020 agricultural census, with a core collected every few years and modules on labour, rural development, machinery, orchards and vineyards. A later regulation on statistics on agricultural input and output covers crop and animal production, prices, nutrients and plant protection products. The Economic Accounts for Agriculture provide agricultural output, intermediate consumption and income consistent with national accounts. The LUCAS survey (Land Use/Cover Area frame Survey) is a point-based area frame survey of land use and cover across the EU. Farm-level accountancy data come from the Farm Accountancy Data Network, being extended into a Farm Sustainability Data Network.
Many countries combine a periodic census with annual surveys and administrative data. Statistics Canada conducts its Census of Agriculture every five years alongside the population census and increasingly uses administrative and remote-sensing data. The Australian Bureau of Statistics runs an agricultural census every five years with surveys in between. Japan's Ministry of Agriculture, Forestry and Fisheries conducts a Census of Agriculture and Forestry every five years. Brazil's IBGE conducted its agricultural census (Censo Agropecuario) for 2017 and runs monthly crop surveys. China has conducted three national agricultural censuses, the third with reference year 2016. Comparing these systems shows common trends: multiple frames, integration with population censuses, registers and remote sensing, and growing attention to farm households, gender and sustainability.
Comparisons suggest priorities for India: a master sampling frame linking land records, census and remote sensing; reducing dependence on manual field-by-field recording through digital crop surveys; integrating household surveys with farm statistics; publishing release calendars and revision analyses; and strengthening the statistical independence of agricultural estimates. Several of these are under way through initiatives such as the digital crop survey under the Agristack programme, but their statistical quality needs independent evaluation.
Statistical agencies differ in how they release market-sensitive agricultural statistics. NASS publishes a calendar a year in advance, releases reports at fixed times, and keeps the forecasts confidential until release. Eurostat publishes release calendars and metadata in a standard format. In India, release calendars exist for many NSO products, and DES publishes advance estimates at announced intervals; publishing revision histories and methodological notes alongside each release improves trust and helps users understand the reliability of the figures.
A student of agricultural statistics must know where each number comes from, how it is defined and how reliable it is, both for exams and for using secondary data correctly in a thesis.
A five-year statistician in DES, a state directorate, NSO, an ICAR institute or FAO compiles land-use, crop, livestock or price statistics, checks consistency between sources, prepares advance estimates and documents methods for users.
A senior official designs or reforms parts of the system: introducing digital crop surveys or remote sensing, revising censuses and surveys to international standards, negotiating data sharing between ministries, and defending the independence and quality of official estimates.
A PhD with long experience evaluates and redesigns statistical systems, advises FAO, the World Bank or national governments under the Global Strategy and 50x2030, and researches integration of surveys, registers and remote sensing.
Institutional arrangements, release schedules and figures change; numbers are rounded illustrations and should be checked against current publications. Details of the post-2023 advance-estimate calendar, the Agristack digital crop survey and the 21st livestock census are described only in general terms. State-level variations in land-record systems are not catalogued. Fishery and horticulture estimation designs are summarised, not specified. Other national systems are sketched, not described in full.
Original practice questions written to the syllabus and pattern of the exams named in each question's examTags. They are not past papers and do not reproduce any board's questions. Every numerical answer was computed with a script; the values needed are stated in the question. Institutional figures are rounded and may change; check current publications.
These notes are free to read here, with no account.
What the Laisha app adds:
Free for ten days, then a subscription. We would rather write that here than have you find it later.