Money Not To Burn

Can paying farmers reduce crop residue burning?

If one is in Delhi in Dec-Jan, they can see a thick layer of smog. Many days, air pollution touches extremely severe and hazardous categories of Air Quality Index (AQI). Crop residue burning in neighboring states of Punjab, Haryana, Uttar Pradesh is considered a big contributor to this pollution. Farmers in this region grow rice between June-October and wheat between November-April. There is a short time window to transition between crops. In this short window, the farmers need to get rid of the crop residue from rice so that they can grow wheat. There are effective ways to get rid of the residue, but they cost money. So, the farmers tend to resort to the cheapest option, which is to burn the residue. In today’s post, we discuss a paper by B. Kelsey JackSeema JayachandranNamrata Kala & Rohini Pande where they study the impact of financial incentives on reducing crop residue burning.

By some estimates, the crop residue burning accounts for 30-40% of city’s winter air pollution. So, while the action seems ‘free’ to the individual farmers, it has negative costs in terms of air pollution. In economics, this concept of your actions having unintended benefits on others is referred to as an externality. An externality can be positive or negative. In this case, farmers burning crop residue is a negative externality on people of Delhi. An example of positive externality will be when I do my singing practice and my wife gets to listen to it. Well, to be honest, that’s also a negative externality only for her. Coming back to the question at hand, there has been a lot of policy interest in encouraging these farmers not to burn their crop residue and use more effective methods like renting a baler which bundles and then removes residue from the plot.

Picture courtesy: The Indian Express,c_limit,f_auto,q_auto:good,fl_progressive:steep/

In general, there are two kinds of financial incentives that are generally used in this case: fine the behavior you want others not to do or to reward the behavior that you want others to do. The governments have tried imposing fines but these farmers are anyway poor, plus imposing fines is also not a popular strategy politically. Instead, now the focus has been on rewarding farmers who actually do not burn their crop residue. This is the topic that the authors study in this paper.

What is the main research question?

  • Does paying farmers when they don’t burn crop residue lead to less burning? These payments are referred to in the paper as “Payments for Ecosystem Services (PES) contracts”. We will also refer to them these payments as PES for this post.
  • The authors make a valid point that a promise to pay farmers after they have not burned may not be ideal for two reasons: one, the farmer may not have the money earlier to go for an alternative treatment of this residue, and second, the farmer may not trust that any payment will be made afterwards. So, the authors also study a different form of PES, which they refer to as upfront PES, in which half of the payment is made beforehand regardless of whether a farmer burns later or not. The upfront PES is trying to tackle both the parts: giving farmers some money to support them in using an alternate method of residue disposal, as well as hoping to earn their trust that money will actually be paid if they don’t burn.

How do the authors study this question?

As we discussed in the previous post, randomized experiment tends to be the best tool available to answer such questions. The authors in this paper also run a randomized experiment in 171 villages in Punjab. Villages were divided into three groups: some villages received no such money transfers, some received no upfront payment but full payment only if they did not burn the residue (authors refer to this as Standard PES), while the rest received some of the payment upfront with the rest being paid if they did not burn the residue (Upfront PES). One key element of this is the amount that the farmers are being paid, because if you pay very high amounts, the compliance might be high in this experiment but it likely won’t be sustainable. Here, the authors consult with Punjab state government and come up at a scalable payment level of 800 Indian rupees (~$10) per acre of land. The cost of using a baler (at the time of study in 2019) is also comparable and costs ~1000 rupees per acre.

Another key element here is evaluating whether a farmer burned the residue or not. In this case, the authors had the staff monitor individual farms in a farm specific window to determine whether the farmer burned the residue or not. This is still possible for farmers who are being paid some money but the farmers in the group without any PES would not allow people to evaluate their farms. Therefore, the authors had to combine satellite data along with their manual burning monitoring data and fit a machine learning model to predict whether a farmer burned or not. We won’t go into all these details here. You can refer to the paper if you are interested to learn more. Once the experiment has been set up, the next step is to compare which group has the smallest burning rate.

What do they find?

The authors find that the upfront payment helps a lot and farmers in the upfront PES were twice as likely to not burn as compared to standard PES. The interesting result was also that the burning with standard PES was not very different from the group which received no payment whatsoever in any case. Overall also, the compliance was pretty low. Even with the upfront PES, only 18.5% farmers complied and did not burn. So, close to 80% farmers received some money upfront but burned either way. But at least, this study shows that if such a scheme were to be implemented, making some payments upfront works better than conditioning the entire payment only afterwards.

In conclusion, the results indicate that there is more to learn about crop residue burning. Even with payments, there was a modest impact on burning. This too, when there was such granular monitoring. The authors admit that their monitoring protocols were not designed for scale. At a state level, there won’t be enough resources to verify whether a farmer burned or not, and thus might lead to even less compliance. The authors also show some optimism in the end though. Given we see a higher compliance with upfront payments, it might indicate that trust in whether payment would be made could be a big factor. If probably the experiment were to be repeated the following year too, more farmers may comply as they would know that the payments were made the previous year. With a statewide program too, once people see payments being made, they may start trusting the system more and compliance could possible increase. But a lot of unknowns remain in this area