On fertiliser subsidy reform

Scroll has a very good series on why the fertiliser subsidy reform plan involving delivering the subsidy directly to farmers as a Direct Benefits Transfer (DBT) by linking land records, soil health cards (SHCs), and Aadhaar number has had to be shelved, even if temporarily. 

The reform objective was to deliver the exact amount of fertilisers to each farmer based on their respective soil nutrient content as captured in the SHC and their Aadhaar-linked land record details. The reform struggled because of poor quality of land records and SHC data and behavioural challenges. Worse still, even the apparently pure technology play of Aadhaar authentication generated surprisingly high failure rates. 

One of the articles writes about the challenges with getting farmers to accept the SHC reports,

Farmers in Krishna district ignored the soil health card information that appeared on the machines. “When it comes to deciding how much fertilisers should be used, we go by our traditional wisdom and by what other farmers are using,” explained Babu Rao, a farmer in Krishna district. “We understand we might be using too much chemicals. But who will be responsible for the loss if we get poor yield by using less fertiliser?”

... Thousands of farmers and dozens of fertiliser retailers in Krishna and West Godavari district did the same – they ignored the machine’s instructions. “We could not deny what farmers asked for,” explained a fertiliser dealer in G Kundur, a block centre in Krishna. “There would have been riots. Our business would have suffered.”

And the inconsistency between the strategy and objective, and targets and resources,

To issue soil health cards, officials test one sample of soil each from a grid of 10 hectares of land in dry areas and two hectares of land in irrigated areas. Experts say this methodology may not produce reliable data for individual farms since the average size of agricultural land holding in India is 1.1 hectares, with 67% of the land holdings measuring less than one hectare. Soil characteristics change from farm to farm depending on the cropping patterns and fertilisers use in those farms. Data on soil characteristics over larger areas can therefore be misleading when used to determine the inputs required on a specific plot of land... In Krishna district, where 33 lakh soil health cards were generated last year from the 95,000 soil samples analysed, said officials. The four soil testing labs in the district had the capacity to test only 21,000 samples a year. The officials claimed private consultants were hired and the labs ran day and night to complete the job.

It is difficult to believe that this experiment could have turned out any different. Consider the requirements. First, the Agriculture Department had to effectively manage the logistics of soil health data collection - maintaining the soil sample identity, transportation and storage of soil samples, their analysis, and its communication back to farmers. Second, the soil analysis had to be an accurate reflection of the soil nutrient status - the sampling had to be representative enough and the tests had to be effective. Finally, the farmers had to have the faith in the results to be willing to accept its findings. Forget the second and third, there are too many moving parts in the management of SHC logistics itself for a weak state (an even more weak Agriculture Department) to have been expected to  deliver with any degree of credibility. 

And we have not even talked about the challenges with land records - mutation and ownership updation in basic land records, sub-division in survey records, disconnect between registration and ownership, lack of accurate tenancy records, predominance of tenancy and so on. Or with Aadhaar validation - poor connectivity, concentrated load on retailers during the sowing season, training of (1,75,619) retailers, maintenance of point of sale terminals, and so on.

The takeaways. One, it is unrealistic to expect weak states to be able to perform herculean tasks covering millions of people with any reasonable degree of quality. Adding unrealistic timelines only makes the likelihood of success even more remote. Two, excessively ambitious and over-engineered solutions are unlikely to work. Practical considerations and related compromises are important. Three, technology solutions cannot paper over fundamental structural deficiencies like poor quality of land records and weak state capacity. Finally, despite the perception to the contrary, technology solutions are not always easily implemented effectively in scale.