Key Details
The Improvement of Crop Statistics (ICS) Scheme, introduced in 1973–74, independently checks the quality of crop-area enumeration and crop-cutting experiments conducted by state authorities. During 2023–24, it operated across 20 states and two Union Territories, covering around 33 crops.
Area of assessment | Finding from 2023–24 | What it reveals |
|---|---|---|
Statistical coverage | Land-record-based systems cover around 86% of the reporting area and sample-based EARAS systems another 9%; about 5% remains without a satisfactory reporting system. | The national system remains institutionally uneven. |
Timeliness of crop inspection | Girdawari was completed on time in only 43% of villages in Early Kharif, 46% in Late Kharif, 48% in Rabi and 33% in Summer. | Delayed field enumeration can weaken early crop-area and production estimates. |
Quality of area recording | Discrepancies in crop or crop-area entries were found in approximately 21–26% of surveyed plots or survey numbers, depending on the season. | High aggregate coverage does not guarantee accurate primary records. |
Crop-cutting procedures | Prescribed procedures were followed in 72% of inspected experiments; essential equipment was unavailable in around 24%. | Yield estimates remain exposed to operational and measurement errors. |
Staffing and supervision | In some states, more than 40% of experiments were conducted by inadequately trained or junior personnel. | Statistical reliability depends heavily on field capacity and supervision. |
Yield estimation | ICS and Crop Estimation Survey estimates have shown greater agreement for most major crops over the past five years, although crop- and state-level differences persist. | Independent checking appears to be improving consistency, but not uniformly. |
Precision of estimates | Standard errors ranged from 0.7% for wheat in Punjab to 19.7% for Kharif jowar in Maharashtra. | The reliability of estimates varies substantially across crop-state combinations. |
Digital transition | Electronic crop-cutting schedules have replaced paper formats, while the Agrisoft application is being developed for app-based supervision of area enumeration. | Digitisation can improve monitoring and standardisation, provided field systems are strengthened alongside it. |
A Large Statistical System with Uneven Foundations
India’s crop-production estimates are built through an extensive combination of village land records, crop inspections, sample surveys and crop-cutting experiments. Under the ICS Scheme, NSO independently checks portions of this work to identify errors in crop-area recording, assess adherence to crop-cutting procedures and compare independently derived yield estimates with those produced by state crop-estimation systems.
The review shows that this quality-assurance architecture is substantial. The number of crop-cutting experiments conducted through the broader estimation system has risen from around 1.73 lakh in 1973–74 to more than 11.6 lakh in 2023–24, while the ICS sample-check mechanism itself covers thousands of villages and tens of thousands of experiments.
However, the underlying reporting arrangements remain fragmented. Most reporting areas are covered through land-record systems, where village officials record crops and land use through periodic Girdawari inspections. Kerala, Odisha and West Bengal, among others, depend substantially on sample-based reporting arrangements. A remaining share of the country continues to rely on less satisfactory conventional methods, particularly in difficult or inadequately surveyed regions.
Field Implementation Remains the Principal Weakness
The report’s most consequential findings concern the quality and timeliness of primary fieldwork.
Only around one-third to one-half of villages completed crop inspection on time across the four major seasonal reporting periods. State-level response also varied sharply: while some states and Union Territories achieved near-complete coverage, response rates fell below 70% in parts of Bihar, Gujarat, Uttar Pradesh and West Bengal.
The checks further identified weaknesses throughout the statistical production chain:
Crop or area-recording discrepancies appeared in roughly one-fifth to one-quarter of inspected survey numbers.
Errors included failure to record an existing crop, recording a crop that had not been sown and incorrect assessment of cultivated area.
Irrigation and seed-variety particulars also differed between primary and supervisory records in a significant proportion of cases.
Many village maps remained usable, but 63% were more than 20 years old, reducing their value for identifying altered plots and land-use boundaries.
Substitution of selected crop-cutting units remained high in several states because crops had already been harvested, sampling information was outdated or cultivators could not be contacted.
Some experiments were conducted without essential measuring and weighing equipment or by personnel who had not received adequate training.
These weaknesses matter because errors introduced during village enumeration or harvesting-stage experiments cannot always be corrected through later scrutiny, software validation or statistical aggregation.
Yield Estimates Are Becoming More Consistent but Reliability Still Varies
A more encouraging pattern emerges from the comparison between ICS yield estimates and Crop Estimation Survey estimates. Over the five agricultural years from 2019–20 to 2023–24, estimates for most major crops have generally moved closer together, suggesting greater methodological consistency between independent checks and the regular estimation system.
The convergence is not uniform. During 2023–24, ICS produced a higher estimate than CES for groundnut, while its estimates were lower for wheat, bajra, ragi, gram, rapeseed and mustard, and Kharif jowar. Some crop-state estimates also carried relatively high standard errors, indicating that closer headline estimates do not necessarily imply equal precision across the system.
The review therefore provides evidence of improving agreement, rather than proof that crop-yield estimation has become uniformly reliable.
Digitisation Can Strengthen - but Not Substitute for - Field Capacity
NSO has replaced paper-based crop-cutting schedules with the AS 2.0 electronic schedule, allowing Regional Offices and State Agricultural Statistics Authorities to collect and transmit experiment data in a standardised format.
It is also developing Agrisoft, an application intended to support app-based supervision of crop-area enumeration and land-use statistics. Such systems can reduce manual entry, strengthen validation, flag missing information and allow supervisors to monitor fieldwork more quickly.
Digitisation, however, cannot by itself resolve late crop inspections, obsolete village maps, inadequate equipment, weak training or poor coordination between state agencies. The next stage of reform must therefore connect digital tools with stronger field institutions and clearer accountability for statistical quality.
What Is Girdawari?
Girdawari is the periodic field inspection conducted by village-level revenue officials (commonly Patwaris) to record what is being cultivated on each agricultural plot during a crop season. These inspections update official land records by documenting the crop sown, cultivated area, irrigation status and other field characteristics.
In India’s crop statistics system, Girdawari forms the foundation for crop area estimation. The information collected is used to prepare crop area statistics, supports agricultural production estimates and provides the sampling frame for several agricultural surveys, including the Timely Reporting Scheme (TRS) and the Improvement of Crop Statistics (ICS) Scheme.
Because many agricultural statistics depend on these field records, delayed or inaccurate Girdawari directly affects the reliability of crop area, production and yield estimates, making timely completion an important indicator of statistical quality.
Policy Relevance
More credible production estimates: Better agreement between ICS and state yield estimates can strengthen confidence in the figures used for agricultural planning, procurement and food-balance assessments.
Earlier policy decisions require timelier fieldwork: Delayed Girdawari and reporting weaken the information available when governments make early decisions on procurement, imports, exports, storage and drought response.
Crop insurance depends on field-level integrity: Errors, substitutions and procedural deviations in crop-cutting experiments can affect yield benchmarks used in assessing insured crop losses.
Uneven state capacity produces uneven statistical quality: National estimates ultimately depend on the personnel, maps, equipment and reporting systems available at the village and state levels. Targeted support is therefore more important than uniform national instructions alone.
Precision should accompany headline estimates: Regular publication of standard errors and other measures of uncertainty would help policymakers distinguish statistically robust estimates from crop-state figures requiring cautious interpretation.
Digital systems need institutional integration: Electronic schedules and app-based supervision can improve validation and traceability, but their value will depend on staff training, updated sampling frames and adoption across state statistical systems.
Climate variability raises the cost of weak data: More frequent weather shocks increase the need for accurate, timely and geographically detailed information on sown area, crop conditions and realised yields.
Follow the Full Report Here: Review of Crop Statistics System in India through the Scheme of Improvement of Crop Statistics, 2023–24