N Chandra Mohan: Poverty of reliable numbers

Govt needs to strengthen machinery for gathering statistics

A wish for 2012 is that the UPA government bestirs itself to strengthen the machinery for gathering statistics, the credibility of which has taken some knocking of late. Earlier, the Reserve Bank of India (RBI) governor had expressed doubts about the reliability of basic statistics: the frequent adjustments to GDP growth estimates, volatility in industrial production indices and inflation data are bound to affect policy formulation and research.

India’s GDP growth is slowing sharply even as another global economic crisis looms over the horizon. But little is known on whether the economy is generating adequate employment opportunities. Unlike the advanced countries where up-to-date information is available on labour market behaviour, this is available in India only with a five-year lag. The last comprehensive survey of the National Sample Survey Organisation (NSSO) pertains to 2009-10.

The 2009-10 numbers, for their part, hardly bolster any confidence in the government’s data gathering capabilities. The NSSO data puzzlingly show niggardly rates of employment growth and declining rates of unemployment! While a lower pace of growth in jobs, if not outright declines, may be expected a year after the global economic crisis of 2008-09, the simultaneous fall in rates of joblessness does appear anomalous.

However, nowhere is the need for improving data collection more acute than in the case of agriculture. India’s agricultural statistics relies on village revenue officials to compile plot-wise data on land use and crop-wise area and estimates of yields based on crop-cutting experiments in sampled villages. This system has deteriorated over time.

The system has failed partly due to the scale of effort involved. Area estimates entail complete enumeration of plots in 120,000 villages — a task that is bound to strain the large number of poorly trained, over burdened and poorly supervised village officials, according to the recent Report of the Expert Committee for Improving Agricultural Statistics. The indiscriminate increase in the number of crop-cutting experiments to generate yield estimates has also made it difficult to ensure that they are done properly.

The Expert Committee recommended putting in place an institutional arrangement that would provide reliable and timely statistics. A smaller sample of 15,000 villages and 90,000 crop-cutting experiments (as against the planned 170,000 experiments and 880,000 actually done, at present) is adequate for reliable numbers. The operations involved must be planned, managed and supervised by a National Crop Statistics Centre in the Ministry of Agriculture.

Another strategy would be to strengthen remote sensing to independently collect primary data on land use and crop data that reduce the scope for human error. However, this technology as yet is far from being a magic bullet to reduce dependence on human agency in gathering data. This was indicated in a study comparing data thrown up by remote sensing and field studies in 12 villages with diverse agro-climatic zones and cropping systems in four states of the country (Andhra, Gujarat, Karnataka and Uttar Pradesh).

Though the estimates of the extent of agricultural land were in agreement in most states, the Expert Committee noted that estimates of total cropped area, area under different crops in different seasons differ from those obtained by the field survey in differing degrees. The Kharif area is close to the survey estimate in two out of the five villages but much higher than the survey estimates in remaining villages. The differences are also wider in identifying specific crops grown on different plots and estimated area under each.

While remote sensing has the capability to produce reliable agricultural statistics, it was felt that more work and experience is needed to exploit its full potential. The recommendation is that it must be viewed as a “complementary system” and not a substitute for conventional methods.

While the suggestions for a National Crop Statistics Centre cannot be postponed, the others that call for addressing the inadequate staffing pattern of the Directorate of Economics and Statistics are of general import. This problem bedevils other departments of the government that generate highly provisional numbers like indices of industrial production and inflation.

When the 2009-10 data became controversial, defensive government officials alluded in the media to faulty data collection of the NSSO; that insufficient probing questions were being asked to capture those willing to work or were seeking or available for work. There were also suggestions that NSSO survey work has been done by contract staff that has affected data quality. Unless these problems are addressed, this poverty of reliable numbers will certainly impact policy formulation and research.

From the Ivory Tower will make research from the academic world accessible to all our readers