Introduction and motivation

Crop residue burning (CRB) refers to the burning of crop residue, following the harvest season. CRB usually takes place in open fields and is used as a strategy by farmers to prepare farms for the next cropping cycle. CRB occurs with seasonal regularity across much of South and East Asia in countries such as India, the Philippines, China, and Thailand. The adverse implications of this practice have been recognized with increasing urgency in recent times. Of note is the substantial increase in air pollution that has significant implications for environmental and public health outcomes [1–4]. The polluted air spreads from its place of origin to neighboring rural and urban areas [5–10]. This form of air pollution has been associated with increases in respiratory illnesses and public health costs [11,12]. In addition to its contribution to air pollution, CRB has been known to negatively affect soil quality by reducing organic matter [13]. This in turn can result in lower farm yields over prolonged periods of time [14]. CRB therefore becomes pertinent from the perspectives of sustainable farmland management practices, the natural environment, as well as public health.

In the case of India, much of the recent research on CRB has focused on the rice crop in particular [3,5,15–21]. Rice is the major summer crop in large parts of the Indo-Gangetic plains. Typically sown during June-July, it is harvested in November-December. Farmers typically do not reap rice residue and choose to burn it because of its limited usability. In contrast to rice, little or no attention has been paid to CRB practices related to wheat which constitutes the second, winter crop in large parts of Northern and Eastern India [21,22]. Similarly, attention has not focused on other parts of India where crops such as rice, sugarcane, and oilseed are also subject to postharvest residue burning; for example, in western and southern India in states such as Maharashtra, Karnataka, and Tamil Nadu [16,21,23]. Wheat is sown during December-January and harvested in April-May. Unlike rice residue that is burned, wheat residue is characteristically reaped post-harvest because of its tradable value as cattle fodder [24]. [25] report that forty-five percent of wheat residue in Punjab is used as cattle fodder. They estimate that 81 and 48 percent of farmers burn rice and wheat residue, repectively. Similarly [26], estimate these numbers to be 40 and 22 percent, repectively. Although wheat residue burning has not been as common as rice straw burning traditionally [27], report that farmers in northern India have increasingly been using CRB as a postharvest practice for wheat as well. A recent examination of satellite data on fire counts in northwest India suggests that wheat residue burning has been on the rise between 2014 and 2020 [22]. It is against this background that our study examines the socioeconomic reasons that drive the practice of CRB for wheat residue in the state of Haryana, India.

Using primary data on farmers’ postharvest residue management practices this study contributes to an emerging evidence base of wheat residue burning in India. The curiosity that wheat CRB occurs at all, even though its residue retains value as cattle fodder, merits closer investigation. We do this firstly by conducting a benefit-cost test to examine whether it is in the farmer’s financial interest to choose CRB instead of non-burning alternatives for wheat residue management. Secondly, we examine the extent to which a farmer’s decision to burn wheat residue is conditional on their CRB decision of the (preceding) rice crop. We explore the possibility of a relationship between the burning of rice and wheat residues since the cultivation of both crops is intensively mechanized with combine harvesters, which is an important reason for choosing CRB.

To our knowledge, this is the first study to draw a link between CRB practices across crops grown in different agricultural seasons. Our study is well-placed to investigate these relationships for the following reasons. First, the agricultural practices in Haryana reflect the typical rice-wheat farming system. This allows us to identify the factors that can explain the reasons for wheat CRB. Second, our analysis relies on primary data of residue management practices collected over two consecutive harvest seasons of rice (November-December in 2018) and wheat (April-May in 2019). Furthermore, the findings from our previous study on the drivers of rice CRB in Haryana [17] indicated that the perception of CRB being a common practice significantly increases CRB, while the perception that it lowers soil quality makes CRB less likely. However, an awareness of its negative environmental effects did not lower its occurrence–suggesting a public goods characteristic of CRB. Given that this study examines the wheat CRB choices of a subset of the same rice farmers in a subsequent harvest season, it allows us to establish how the decision to burn wheat residue is related to burning rice residue in the previous season.

The rest of the paper is organized as follows. In the next section, we describe the background of CRB in India to provide context to our study. We follow this with a description of our sample selection and empirical strategy. In the Methodology section, we lay out the components for a benefit-cost test of wheat CRB and follow that with a description of a regression framework with a set of testable hypotheses. The Results section presents key findings. This is followed by a discussion section with pertinent policy recommendations and a conclusion section.

Background of CRB in India

The earliest documented cases of crop residue burning (CRB) in India were in the mid-1980s [28]. Farming in India began to be intensely mechanized in the 1970s from the introduction of combine harvesters [29]. The use of machinery spread across the north Indian states of Punjab, Haryana, and Uttar Pradesh when farmers switched to an annual rice-wheat cropping system. Farmers began to use combine harvesters in conjunction with high-yielding varieties of rice and wheat crops. One consequence of adopting high yield crops and intensifying farm mechanization was that vast quantities of postharvest crop residue began to be generated in situ [4]. In northern India the high yield rice crop’s residue has little use to farmers but by contrast wheat residue is used as cattle fodder. While both crops’ residues get burned rice CRB is more prevalent than wheat CRB [25,26]. Each year rice or paddy postharvest residue is burned during October-November and wheat postharvest residue is burned during April-May [21]. The time during which rice residue is burned is a season in northern India when the winter air is relatively stagnant–leading to the formation of heavy smog which does not get easily dispersed (Source: https://weather.com/en-IN/india/science/news/2018-10-30-why-do-pollution-levels-skyrocket-during-winter). In the summer months of April-May, the burning of wheat residue is less evident since the air is not stagnant like in winter. This is plausibly one reason why wheat CRB does not garner as much attention as rice CRB.

Apart from rice residue being of little use to farmers they tend to burn it because of a quick turnaround required for the subsequent wheat crop; the best time to sow wheat runs from November through December–roughly two weeks after rice is harvested [4,27]. The earlier non-high yield varieties of rice and wheat, which are not grown much anymore in northern India, used to have cropping cycles that lasted longer than the currently prevalent high yield varieties that have shorter rotations of 120 days. Farmers would previously produce only one crop–either wheat or rice–in a farming year [13]. With the adoption of short rotation high yield varieties, the yearly trend shifted to rice being grown between June and October and wheat being grown between November and March [4]. With these crop rotations little time is left between the rice harvest in October-November and the sowing of wheat in November-December. Delays in wheat sowing can adversely affect its harvest quality.

The conundrum is that with a gap of 46 days between wheat residue removal and the transplanting of paddy there are no binding constraints on farmers to burn wheat residue [13]. Even with wheat residue being useful as cattle fodder and farmers not being time constrained to sow the subsequent rice crop, there is increasing evidence of wheat residue burning. Studies in the literature have reported on other reasons for CRB in general. These include the seasonal shortages of migrant farm labor [21], the fact that it is seen as a quick disposal method by farmers, and that residue burning is considerably less expensive than its machine removal [4,15,17,27,30,31].

Data

The primary data for this analysis comes from the district of Karnal in Haryana state, northern India. The district of Karnal is a homogenous agroclimatic zone in northern Haryana and highly productive agriculturally [32]. There are 78,265 farmland holdings in Karnal, covering 388,827 hectares, of which 187,012 hectares are under wheat cultivation [33]. In 2018 we surveyed 1,230 households across 12 villages in Karnal on rice residue management practices [17]. For the current study, we selected a random sub-sample from 8 of the original 12 villages to be resurveyed in 2019, specifically on their wheat CRB practices. Our target was 40 farmers per village. During the selection process, we excluded farmers who only grew the non-coarse varieties of rice (i.e., basmati) or non-paddy crops since the residues of these crops typically are not burnt. This process resulted in between 35 to 40 farmers for each village for a total of 301 farmers, who grew coarse varieties of rice among other things, participating in the 2019 survey.

The random sampling for this research was designed to answer the questions of whether choosing CRB made economic sense for wheat farmers, and if there is a relationship in the residue management choices between rice and wheat. The 2019 survey took place immediately after the wheat crop harvesting season in the first half of May 2019. Almost all farmers stated that the harvest of the wheat crop takes place during the last week of April and the first week of May. Each survey respondent was either the household head or the principal household person making agricultural decisions.

Each survey collected information on household demographic and socioeconomic information. Additionally, farmers were asked about their wheat residue management practices. The questions included what their options of wheat residue management were, the reasons for choosing to burn wheat residue, their use of residue if not burned, their perceptions on whether residue burning is a common practice in the village, and their perceived effects of residue burning on soil quality and the natural environment. All questionnaires were translated from English to Hindi, see the English version in S1 Appendix.

Methodology

This paper analyzes crop residue burning (CRB) of wheat residue at three levels. To begin with, we analyze the various factors that influence a farmers’ decision to burn wheat residue. We also compare socioeconomic characteristics of households that did and didn’t burn wheat residue. Furthermore, to understand if there is a link in the CRB decision between crops, we test for the correlation between households that adopted CRB of wheat with that of rice. In addition to this, we undertake two additional sets of analyses- testing for the benefits and costs of CRB and a regression framework. In order to comply with the Institutional Review Board requirement to interview 301 farmers, we got the study approved by the IRB Board at the American University of Sharjah as per protocol #17-406RN with exemption under 45 CFR 46.101(b)(5). Oral consent was recorded for each participant. Extensive meetings were also held with village-level governing bodies (Panchayat) before the start of survey activities to share information about the study objectives.

Results

Descriptive statistics

In Table 1 we summarize the residue management practices by village. We note that almost all arable land available for wheat cultivation is used to grow either the soft or coarse varieties of wheat. Table 1 shows that many farmers in each village adopt burning as a means of wheat residue removal. The percentages of farmers employing this method in the last five years range from 17.95 to 40.54 percent. Overall, a little less than a third of the respondents reports the burning of wheat residue in the last five years, and about two-thirds deny it. Among those who did not use CRB, more than three-quarters reported they used only machine to remove the residue.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Wheat residue management practices by village.

https://doi.org/10.1371/journal.pone.0296059.t001

In Table 2, we compare socioeconomics and other aspects among these two groups of farmers–those who undertook wheat CRB (Yes) and those who did not (No). Table 2 shows that the average number of farm assets owned, and the average total cultivated land are significantly higher amongst farmers who burn wheat residue than those who reportedly do not burn it. Since farm assets and land are proxies for farmer wealth, this implies that those who burn wheat residue are presumably wealthier.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Socioeconomic characteristics by wheat residue burning practice.

https://doi.org/10.1371/journal.pone.0296059.t002

Table 2 also shows that a greater proportion of farmers who use CRB perceived CRB as a common practice than those who did not use CRB. This suggests that farmers are more likely to use CRB if they consider it to be a common practice in their village. In contrast, a greater proportion of farmers who did not burn wheat were aware of CRB’s negative implications for soil health and environmental outcomes. That is, the awareness of the negative environmental effects of burning and the perception of its adverse effect on soil quality appears to discourage farmers from burning residue. In Table 2, we also see that the average cost of manually removing the wheat residue and the average market value of the collected residue are not significantly different between the farmers who burn wheat residue and the ones who do not. However, the two groups relied on different methods of residue removal with the ‘yes’ group relied more on mechanical removal of residue while the ‘no’ group relied relatively more on manual removal.

Linking CRB for wheat with that for rice

Table 2 also indicates a significant correlation between the two decisions of burning wheat residue and rice residue in recent years. Nearly 80% of the subset of farmers that undertook CRB for wheat also undertook CRB of rice in the previous season. In contrast less than 60% of farmers in the ‘No’ group burned rice residue in the previous season.

These findings related to CRB of rice and the effects of farmer’s perceptions related to the common practice in the village, the effect of residue burning on soil quality, and the environmental impact of their wheat residue burning decision are also mirrored in Fig 1. The top right panel of Fig 1 presents the distribution of farmers in terms of their wheat residue burning behavior as related to their rice residue burning decisions over the last five years. It clearly indicates that farmers who did not burn paddy residue in the last five years are less likely to burn wheat residue as opposed to those who did burn paddy residue in the last five years or those who were unwilling to answer this question. This consistency of burning behavior across two crops may arise from underlying factors that similarly affect both decisions. We explore this further in the regression results section.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Percentage of wheat residue burning based on farmer perceptions and paddy residue burning.

https://doi.org/10.1371/journal.pone.0296059.g001

Although we see that burning of rice residue in one season is correlated with CRB for wheat in the following season, it appears that the reasons for CRB of the two crops are not the same (Table 3). While nearly 60% of farmers cited cost-effectiveness as the most important reason for choosing CRB for rice residue management, less than one-fifth of farmers considered it to be so when it comes to CRB for wheat. For the latter, 77.08 percent of respondents claim that the most important reason for choosing CRB is the use of a combine harvester. As alluded to in the section on Background of CRB in India, combine harvesters were a part of the intensive farm mechanization process since the 1970s, which consequently led to vast quantities of postharvest crop residue being generated. The high cost of managing wheat residue without burning it is the second-most important reason. This is akin to findings in [20] wherein the upfront cost of residue harvesting machinery is an important barrier to adopting environmentally friendly alternatives among Indian rice and wheat farmers. Similarly, [19] report that limitations to farmer’s resources and information hinder the adoption of cleaner production strategies. The lack of residue usability is seen as a reason for choosing CRB by 12.96 percent of paddy farmers as opposed to a smaller 2.66 percent amongst wheat farmers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Primary reasons for choosing CRB.

https://doi.org/10.1371/journal.pone.0296059.t003

Regression results

Table 4 reports five probit model results using Eq 1 as described in the Methodology section. In Models 1 and 3 we exclude 36 farmers who were not willing to answer whether they have used CRB after harvesting paddy. We include them in Models 2 and 4 with a binary variable indicating whether they were willing to answer or not. We exclude two outliers with land size greater than 50 acres. Models 3–5 include the probability weights of household populations in each village. The number of wheat-growing households are sourced from the village panchayat lists. Model 5 includes two additional covariates. We exclude 35 farmers who report that they do not know the price of wheat residue. All the five models yield remarkably similar results. We discuss seven salient findings.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Regression coefficients of determinants of wheat residue burning using probit.

(Binary dependent variable: choosing CRB in last five years–Yes = 1, No = 0).

https://doi.org/10.1371/journal.pone.0296059.t004

We find firstly that our results support Hypothesis 1; i.e., farmers who believe that CRB is a common practice are more likely to choose CRB than those who do not believe so (p-value < 0.01 in all models). Second, the perception of the effects of CRB on soil quality has a significant impact on farmer behavior. Farmers who think that CRB diminishes soil quality are significantly less likely to choose CRB than those who think it improves soil quality or are uncertain about its effect. This result suggests that farmers do internalize their associated costs and benefits of CRB. Third, farmer’s awareness of the negative environmental impact of CRB has a negative sign as one would expect; the effect however is statistically insignificant. Thus, farmers do not substantially internalize the external social costs of CRB in their private decisions. These three findings are consistent with our previous findings amongst a larger set of farmers with regards to their rice residue management practices [17].

Fourth, amongst the various socioeconomic characteristics, education is the only variable in which we observe any statistical significance. Farmers with higher education are more likely to adopt CRB (p-value = 0.029 for Model 1, 0.093 for Model 2, and not significant in Models 3 and 4). The coefficient of land size is positive as expected but statistically insignificant. [17] report that paddy farmers with larger land sizes are significantly more likely to use CRB. The coefficient of the number of machines owned is also positive but not statistically significant since only 3 farmers reported owning a harvester. Livestock ownership has a negative and insignificant coefficient.

Fifth, incorporating a probability weight of household populations in each village–as done in Models 3 and 4 –results in qualitatively similar findings to Models 1 and 2, which exclude these weights. This result indicates a zero bias of population representation in the regression results.

Sixth, for those farmers who reported earlier that they burned rice residue, and even more so for those farmers who were unwilling to answer this question, we note that both groups significantly more likely to burn wheat residue as compared to farmers who did not burn rice residue previously. This result supports Hypothesis 2 –that CRB decisions for rice and wheat are potentially linked, or that there is a spillover effect from rice CRB to wheat CRB.

The test for endogeneity in our regression framework when using rice CRB as an explanatory variable relies on the p-values from the Wald-test. Using rice residue burning being a common practice as an instrument yields the p-values from the Wald-test of exogeneity as 0.32 and 0.57 –when including and excluding probability weights of household populations, respectively. These non-significant p-values imply that we cannot reject the null hypothesis of exogeneity. This suggests that not including an instrumental variable in the regression framework is reasonable. These instrumental variable regression results are available upon request.

Lastly, the coefficients of the cost of wheat residue removal using machinery per acre and the price of wheat residue collected per acre in Model 5 are not statistically significant. We also replace the cost of residue removal using machinery with labor as well as use the net benefit (the difference between the price of residue collected and the cost of residue removal per acre) to replace both variables; none of these coefficients is statistically significant. We noted from the benefit-cost analysis that a high cost of transporting crop residue to fodder markets limits farmers from adopting environmentally friendly ways of residue management.

In Table 5, we report the average marginal effects from the probit regressions results of Models 3, 4, and 5. The results of all the models are similar, and there are three variables where their average marginal effects are statistically significant. First, farmers who believe that CRB is a common practice are 25 to 26 percent more likely to choose CRB than do those who do not believe so (p-value < 0.001). Second, farmers who believe that using CRB diminishes soil quality are 23 to 26 percent less likely to choose it as compared to other farmers who believe otherwise (p-value < 0.001). Finally, those farmers who previously chose CRB for rice residue management, and those farmers who were not willing to reveal their CRB choices, are 19.4 to 21 percent and 15.6 to 17.5 percent more likely, respectively, to choose CRB for wheat than farmers who did not previously employ CRB for rice. The average marginal effects of other control variables–socioeconomic characteristics, benefit of selling residue, and cost of manual residue removal–are not statistically significant; suggesting that these factors are not major determinants of wheat CRB.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Average marginal effects on wheat residue burning (using Models 3, 4, and 5).

https://doi.org/10.1371/journal.pone.0296059.t005

Discussion

This study has taken a closer look at the reasons behind crop residue burning (CRB) of wheat. Consistent with our previous findings concerning rice CRB [17], we infer that the perception of residue burning being a common practice significantly increases wheat CRB, while the perception that residue burning lowers soil quality makes the practice less likely. We similarly find no evidence of environmental awareness lowering residue burning, thereby indicating a public goods characteristic. On tracking the residue management choices of rice-wheat farmers over two consecutive cropping seasons, the current study infers that 31 percent of farmers burn their wheat residue and that they are 21 percent more likely to do so if they had previously burned rice residue. We also determine that 92 percent of the farmers use some portion of wheat residue as cattle fodder, 52 percent reincorporate some of it into the soil, and 39 percent choose to sell some portion of it.

The benefit-cost analysis suggests that wheat residue attains a potentially positive value net of residue collection costs–more so for machine collection of residues rather than manual collection. While farmers in Haryana are known to use wheat residue as fodder, it is seen to be the case that several of them burn it even when there is little compulsion to do so. This paper has evaluated evidence relating to wheat CRB occurring notwithstanding a positive net benefit of selling it. This counterintuitive result has been examined in the context of rice CRB decisions possibly influencing the same farmers’ choices concerning wheat CRB in a subsequent cropping season. Farmers appear to experience a spillover from their previous rice residue burning practices even though it may not be their most rational choice. This suggests that policies designed to curb the use of CRB as a farm management strategy should not focus on specific crops in isolation. Rather, it is critical to inform the design of such policies by looking at the agricultural calendar as a whole and realizing that farmers’ decisions related to the management of one crop can have spillover effects for a second crop.

Meeting the challenge of crop residue burning requires carefully targeted policy interventions to work effectively, thereby making it imperative that policy interventions are designed in accordance with how people behave [37,38]. Accordingly, the identification of pertinent behavioral factors driving crop residue burning could bolster the design of effective policies to reduce its occurrence. Wheat CRB was higher when farmers perceived it to be a common method of managing crop residue. In this respect, policymakers could use informational campaigns or nudges to encourage individuals to conform to social norms of pro-environmental behaviors. For instance, [39] provide a meta-analysis of field experiments that rely on social norms in promoting pro-environmental choices; [18] consider the same in a crop residue burning context.

Farmers could potentially gain from selling their surplus wheat residue. However, limiting factors remain for the farmers–such as an inadequate access to residue markets, uncertainty about the marketability of fodder, variability in fodder prices, and high market transportation costs. Ensuring that farmers have access to well- functioning markets for wheat residue could avert the scaling-up of wheat residue burning amongst Indian wheat growers. Such markets could prove financially remunerative to farmers by increasing their perceived net benefit of pro-environmental practices. An avenue worth exploring here might be to subsidize the cost of transporting residue to markets. The persistence of pro-environmental choices could arise with assured financial gain for the farmers. [16] report that the economic benefits of sustainable alternative uses of crop residue should be made clear to farmers for policies to effectively reduce CRB. In a similar vein, [40] finds that farmers persist with environmentally-friendly tillage decisions based even on temporary financial incentives.

Conclusion

The burning of postharvest crop residue (CRB) has come to acquire significance from the perspectives of sustainable farmland management practices, the natural environment, and public health in developing economies with large agrarian sectors. Addressing the drivers of farm management practices by agricultural stakeholders assumes relevance from the standpoint of sustainable land use. India, with its large agrarian sector, regularly witnesses postharvest crop residue burning and bears substantial public health and environmental costs. While rice residue burning worsens air quality every winter and continues to receive a lion’s share of the attention in the literature, wheat residue burning on the other hand, although known to occur every summer, has not received much attention. This paper addresses this gap by examining the drivers of wheat residue burning and drawing a relationship to CRB of rice.

With scarce attention paid to the phenomenon of wheat residue burning in India, our paper is the first to investigate its drivers. One reason identified for residue burning is the intensive mechanization of agriculture since the 1970s such as the adoption of combine harvesters for the rice-wheat dual cropping system in northern India. While rice residue burning occurs regularly on a wide scale every winter, the novelty of this paper is that it turns the lens on the possibility that farmers shadow their rice residue management practices from prior seasons when it comes to the wheat harvest.

We note that we can only cautiously extrapolate or generalize our findings to the broader region of this part of India since this study is limited by the size of its sampled farmers and the choice of a single district in the state of Haryana. Furthermore, our data pertains to a single crop harvest season, so we cannot discern long-term trends, for which the sampled households would need to be repeatedly surveyed on their CRB decisions over successive seasons. Despite the data-driven limitations, this study does identify an emerging yet important gap in the knowledge of CRB occurrences. Crop residue burning is topical to several developing countries and this paper contributes to a growing narrative on agricultural residue management practices that have wider environmental and public health consequences. Future research calls for the conceptualization of targeted policies that foster pro-environmental agricultural practices.

Supporting information

Acknowledgments

We acknowledge the receipt of a faculty research grant (#FRG17-R-49) and the Open Access Program (#OAPSBA-1210-B00016) from the American University of Sharjah. This paper represents the opinions of the authors and does not mean to represent the position or opinions of the American University of Sharjah. We thank Probe and Search India for their professional survey work and Soumya Gupta for her comments on the initial draft of this paper. We also thank seminar participants at the American University of Sharjah and participants at the Southern Economics Association Meetings in 2022 for their comments and suggestions.