Data collection could help India tackle its food inflation problem
The recent spike in Indian inflation is a reminder of how food prices are notoriously volatile. From weather and cultivation patterns to consumer tastes and incomes, the drivers of price fluctuations—and their inter-linkages—are too vast to be captured in one framework. But an important factor behind these fluctuations is the persistent swings from oversupply to undersupply. Especially in the absence of adequate cold-storage facilities, relatively minor changes in supply can have large effects on the prices of perishables.
Why are there such fluctuations in supply? Can we minimize them?
Erratic weather drives a large part of supply fluctuations. However, there is another problem that is illustrated through the Cobweb model, first proposed by the agricultural economist Mordecai Ezekiel in 1938. Its major argument is that farmers have miscalibrated expectations. Farmers base their expectations of future prices on what they see in the spot markets at the time of sowing. For example, when farmers see high prices of onions, they grow more of these, thus setting the stage for overproduction. It works the other way around. The result is a “coordination failure”.
Price forecasts drive farmers’ production decisions. The intervention, at an abstract level, is then rather obvious—provide farmers with better forecasts.
But can the government do it? If yes, how?
This piece is an attempt towards a specific proposal.
In this era of technology and smartphones, governments can develop the infrastructure to gather real-time data, even if not perfectly accurate, on the area under cultivation for a few important crops, with their expected harvesting dates. One way to do this could be to develop a smartphone app for farmers to enter this information after sowing and get modest monetary incentives. The mechanism is relatively straight- forward. Since not all farmers will make sowing decisions at the same instant, if the app shows that several farmers are sowing, say, onions, its price is likely to be low due to oversupply. Therefore, the app could advise farmers who make sowing decisions, even if slightly late, in favour of another crop for a better price. This also ensures the early movers, as the farmers who make sowing decisions later have no incentive to grow onions if too many farmers before them have already done so. We want to emphasize that we are not suggesting a socialist-era relic of centralized production planning. Rather, price forecasts themselves will serve this purpose.
We are not the first ones to suggest providing farmers such price information. There is extensive literature in economics on this. Several studies have been done. The evidence is mixed. The key difference in our proposal is that we are advocating an intervention at the production stage, whereas in most field experiments we are aware of study interventions at the last stage, i.e. around harvest time.
One may call the whole idea far-fetched on three grounds. First, can farmers use smartphones to provide such data to the government? Second, can farmers diversify so easily? And third, do we have algorithms to forecast future prices? These are valid concerns.
On the first, there is a somewhat patronizing tendency of questioning the ability of farmers to use smartphones. With cheap data, these devices have already penetrated deep into the hinterlands. Perhaps we may need to develop apps based in local languages. But, if such an app were to be developed, and if the government could provide modest monetary incentives for people to report what they sowed, it is hard to see why farmers would not comply. Naysayers will ask, “What if people cheat and misreport?” This would be the most crucial aspect of the design. Possibly, with mechanisms such as random audits and the monitoring of eventual sales, it is possible to prevent malpractices.
The ability of farmers to diversify their crops has been a longstanding challenge. But having reliable forecasts well ahead in time may just prompt them to diversify. Finally, the above two are probably questions to be studied via a field experiment.
The last concern—the ability to provide reliable forecasts—there has been considerable progress using techniques such as machine learning. Recent advances in time series forecasting and machine learning have led to vast improvements in algorithms that forecast prices. In fact, some also have mechanisms to detect anomalies like hoarding (e.g. a paper by Lovish Madan and co-authors. published in ACM Compass). To be fair, we are not aware of algorithms that provide forecasts far out, but they also do not have information on areas under cultivation, which we believe is now workable to get, as mentioned before.
Devising a system like what we have described above can play a meaningful part in reducing the price volatility of farm produce and offering a policy option to mitigate risks beyond the traditional focus on support prices or forward markets or contract farming. It could also help governments manage the delicate balancing act between protecting the interests of food producers and consumers—a central challenge of Indian political economy. A climate controlled food cold storage is the key to preserve food and fresh produce that requires to maintain customised climatic conditions such as temperature and humidity according to the product requirement. Gubba Food Cold Storage’s robust refrigeration system and technology maintain the condition of the products.