Introduction
The increasing demand for protein has led to rapid growth of the livestock industry and the intensification of livestock production (Zhuang et al., 2020; Briukhanov et al., 2021). Animal herd growth in both developed and developing countries generates increasing quantities of animal waste, most of which consists of manure (Kuhn et al., 2020). On one hand, manure is a livestock residue that can cause high environmental burdens if improperly managed (Joshi and Wang, 2018; Yalcinkaya, 2020; Li et al., 2021). To mitigate the environmental cost, livestock farms are required to store and properly manage the excess manure (Yalcinkaya, 2020). On the other hand, it is also a valuable resource as a source of nutrients (for example, nitrogen and potassium) for crops (Loyon, 2017; Mkuhlani et al., 2020), fish and input for biogas production (Welsh et al., 2019; Ramírez-Islas et al., 2020).
The framework of livestock production is changing from small household crop-livestock farms to specialized and intensive livestock farms (Sorathiya et al., 2014; Huang et al., 2016). And global trends point toward larger, industrialized and regionally concentrated animal operations (Mccann et al., 2005; Havet et al., 2014). Taking China as an example, large livestock farms reached around 58% in 2017 and current trends suggest they will exceed 70% in 2025 (General Office of the State Council, 2020). Larger livestock farms can reduce production costs through economies of scale (Duvaleix-Tréguer and Gaigné, 2016), which benefits both consumers and agricultural producers. However, large farms have a greater potential to cause environmental problems than smaller farms (Aguirre-Villegas and Larson, 2017; Du et al., 2020). These larger livestock farms can cause excess manure accumulation in a concentrated area which can aggravate negative environmental impacts such as nutrient pollution and greenhouse gases emissions (Briukhanov et al., 2021). For example, increased concentration of livestock farming accumulates more livestock manure in the proximity of livestock farm, resulting in the increase of livestock manure concentration and loadings, which adds difficulty to its treatment (Camilleri-Rumbau et al., 2021; Hills et al., 2021). The primary driving force is the incoordination of crop-livestock systems (Wang et al., 2021a). Nonetheless, increasing farm size is a common and irreversible trend in the development of the global livestock industry, making manure disposal a leading challenge in the expansion of livestock farms.
To cope with the challenge of manure disposal, recycling manure for its nutrient or biogas value is becoming a common trend in the livestock industry (Li et al., 2021; Wang et al., 2021a). When taking into account societal interests, manure recycling is a beneficial technique used by farmers to reduce pollution from livestock farming and thus protect the environment. Considering the interests of the individual farmers, manure recycling generates an economic incentive for large farms to treat manure. Instead of merely complying with environmental policies requiring them to do so, these farmers can create another stream of profit by recycling their manure.
Manure can be recycled either onsite or offsite by various agents, depending on which modes of manure disposal a farmer chooses. Common forms of manure disposal modes include self-use, selling, giving away and donation (Shu et al., 2017; Wang et al., 2021a). Previous research shows that manure disposal modes vary significantly with farm size (Aguirre-Villegas and Larson, 2017; Shu et al., 2017; Pan et al., 2021). Under the context of rapid farm size expansion and manure accumulation, it is important to explore the relationship between farm size and farmers' choice of manure disposal modes. And understanding this relationship can help policymakers create better environmental regulations and livestock industry policies.
Material and methods
Research hypothesis
Larger-scale livestock farms create the opportunity to adopt more diversified methods of recycling manure onsite. Hog farming today provides a good example of these more diversified recycling methods. With the expansion of hog farm size, although the proportion of manure returning to cropland is decreasing, a bigger portion of hog manure is being used by farmers for other recycling methods, such as converting manure to aquatic feed or for biogas production (Qiu et al., 2012; Pan, 2016). Specifically, larger farm size correlates with a reduction in the ratio of manure being stockpiled or land applied. It also increases onsite biogas production and organic fertilizer production, as well as selling to offsite farms (Kong et al., 2016; Pan, 2016). Another study based on hog farms found a significant negative effect of farm size on self-use of manure, but no significant impact on other manure disposal modes, such as giving manure away or selling to neighboring farmers and manure dealers (Shu et al., 2017). Xu et al. (2016) on the other hand found that farm size had no significant effect on farmers' choice among three modes of manure disposal: stockpiling, self-use (e.g., for land application and biogas production) and selling. This inconsistency in the findings may be related to the selection of research objects, model construction and variable settings.
The effect of farm size on manure disposal modes can be partially explained by the Environmental Kuznets Curve. The Environmental Kuznets Curve summarizes an inverted U-shaped relationship between the levels of economic development and environmental pollution (Grossman and Krueger, 1995). This inverted U-shaped relationship has also been shown to exist in livestock production, demonstrating the relationship between farm size and pollution indices including total nitrogen surplus and pollution concentration (Yao, 2015; An et al., 2018). Similarly, when positive environmental indices are adopted, there is a U-shaped relationship between farm size and environmental efficiency (Wang et al., 2019), indicating that both small and large farms generate less pollution than medium farms in livestock production (Zhou, 2011). A similar non-linear relationship may exist between farm size and manure disposal modes. In the hog sector specifically, previous research indicates that medium-sized farms have the lowest ratio of manure self-use (i.e., land application), compared to small- and large-size farms (Rao and Zhang, 2018; Pan et al., 2021). This is because land acquisition can be difficult in countries and regions lacking land trading and leasing markets. These difficulties in land acquisition tend to limit the cropland area necessary to meet increased manure needs following livestock herd expansion. As a result, expansion in livestock farms may increase manure giving away to neighboring farmers (Shu and Qiao, 2019; Wang et al., 2021a), leading to a decrease in the amount of manure self-used onsite. When herd size keeps increasing, it is typically more efficient to invest in specialized equipment to process manure for value-added products such as biogas and organic fertilizer, causing a rebound in the self-use of manure and a decline in giving away (Shu and Qiao, 2019; Pan et al., 2021).
In addition to farm size, existing research shows that farmer' choice of manure disposal modes is influenced by a set of common factors, which can be divided into five categories: (1) individual or family characteristics, like gender, age, identity, education level, per capita sown area in a family, per capita net income and the proportion of non-agricultural labor force (He et al., 2016a, 2016b; Si et al., 2019a; Tan et al., 2020); (2) social and economic characteristics, such as population density around a farm, the number of enterprises in a village, the distance from a residential area, a township government or a county, access to loan and the type of residence (Qiu et al., 2012; Triguero et al., 2016; Araya, 2020; Wang et al., 2021b); (3) farm characteristics including farm income, technical training received, time in operation, the participation in agricultural cooperatives and vertical integration level (Pan et al., 2019; Huong et al., 2020; Yao and Zhang, 2021); (4) environmental policy factors, such as pollution penalties and subsidy incentives (Pan, 2016; Si et al., 2019b; Li et al., 2020); and (5) other factors, like the qualitative characteristic of manure, manure value, seasonality and manure price (Mcroberts et al., 2018; Hansen, 2019).
Although manure disposal is important for all livestock species, cattle produce the largest amount of manure per unit of time, and they have an excretion coefficient much higher than other animals (Wang et al., 2021a). Because of the bigger impact that cattle have on the environment (Vries et al., 2015; Pexas et al., 2020), we choose beef cattle as the research subject in this paper. Using primary data from a field survey of beef cattle farmers in China, this paper aims to examine the impact of farm size on farmers' choice of manure disposal modes. Based on the survey data, the manure disposal modes of cattle farmers are categorized into three forms: (1) selling to neighboring farmers, (2) giving away to neighboring farmers and (3) self-use. Based on the literature review, we propose two hypotheses in this paper:
Hypothesis H1: Farm size affects manure disposal, and this effect varies by disposal modes. On average, the larger the farm size, the less likely the farm is to give away manure or choose self-use, and the more likely the farm is to sell its manure.
Hypothesis H2: Farm size has a non-linear effect on manure disposal modes. As farm size increases, the likelihood of farmers choosing to self-use or sell manure first decreases and then increases. On the other hand, the probability of giving away manure first increases and then decreases with size.
Model building
We estimate the relationship between farm size and manure disposal modes through regression analysis. The three dependent variables are the proportion of manure that farms give away, sell or keep for self-use. Each dependent variable is regressed to a set of explanatory variables including the farm size and other characteristics identified through literature. Notably, by the design of research, the dependent variables satisfy two conditions for each sample observation: (1) the add-up constraint that the three proportions sum to one and (2) each proportion value ranges between zero and one.
We employ three estimation methods in the regression analysis in light of these constraints. The primary method we rely on is the constrained singular Seemingly Unrelated Regression (SUR) model (Haupt and Oberhofer, 2000). The SUR model assumes that there are n equations (i.e., n explained variables), each equation has m observations (i.e., the sample size is m and m > n), and the i-th equation has Ki explanatory variables (Ki may be the same or different) (Chen, 2016). Then the i-th equation can be expressed as:
where y i is the dependent variable with i = {1, 2, 3} denoting the three modes of manure disposal, X Ki is a set of explanatory variables, β i is the vector of parameters to be estimated, and ɛi is the residual term.
The SUR model stacks the n equations to jointly estimate as a system of
The covariance matrix Ω of these equations is expressed as:
The advantage of the SUR model is that it can capture the cross correlations among residuals in estimating the covariance matrix Ω. When the off-diagonal terms of Ω are non-zero, accounting the cross correlations can improve the estimation efficiency. For the three equations on the manure disposal modes, because there are overlaps in the explanatory variables, there is a high likelihood that residual terms are correlated across equations. The SUR model jointly estimates all three equations to improve the estimation efficiency.
Because the sum of the dependent variables of the three equations is equal to one for each sample observation, an add-up constraint of y 1 + y 2 + y 3 = 1 needs to be imposed on the estimation following the Constrained Singular SUR model of Haupt and Oberhofer (2000). When the SUR model estimates the three equations simultaneously, the covariance matrix of the residual term is singular. Therefore, the equation for the dependent variable y 1 is removed, and only the remaining two equations are estimated by the SUR model. The parameters in the equation for y 1 are calculated based on the regression results and the add-up constraint between the equations.
To provide a reference for the constrained singular SUR estimation, we run regressions with the same data using the Tobit model with upper and lower censoring limits. The Tobit model estimates the three equations individually instead of jointly. The censoring upper and lower limits between zero and one are specified for the dependent variable y i in each equation. Although this specification accommodates the distributional range for the dependent variables, the estimation strategy cannot guarantee the add-up constraint of y 1 + y 2 + y 3 = 1 being met. For this reason, the results of the Tobit model are only used for comparison purpose.
In addition, we employ the multivariate Tobit model (MV Tobit) for regressions as a robustness check of the SUR estimation. The MV Tobit model jointly estimates the equations to improve efficiency as in the SUR framework (Cornick et al., 1994). However, due to the singularity condition of dependent variables, we can only estimate two of the three equations in the Tobit model. The major purpose of the MV Tobit estimation is to verify the existence of residual correlations across equations, which supports the use of constrained singular SUR model.
Data source
The data used in this analysis is from a primary field survey conducted in Henan Province of China, a major producing area of beef cattle in China. Henan is located in the Central Plain region of China and accounts for 27.39% of China's total beef output in 2018 (National Bureau of Statistics of China, 2019). We select four prefecture-level cities in Henan province to perform the field survey. These four cities have large cattle production, hosting 35.8% of total breeding stock in Henan province in 2016 (Survey Office of the National Bureau of Statistics in Henan, 2017). The field survey is conducted in two steps. First, in July 2018, the research team selected three counties (Biyang, Queshan and Pingyi) in Zhumadian City, Henan Province for pre-investigation. The survey questionnaire was revised and improved based on the feedback from the pre-investigation surveys. Second, in August 2018, the formal survey was conducted in three prefecture-level cities. In each city, three large beef cattle breeding (livestock) counties were selected, and 25 beef cattle farmers were randomly surveyed in each county. The locations of these cities and counties are shown in Figure 1. A total of 289 questionnaires were obtained. In the post-survey quality check, 276 valid questionnaires were obtained after eliminating the invalid ones which were self-contradictory or missing key variables, with an effective rate of 95.5%.
Fig. 1. Maps of study site.
Variable description
Table 1 presents the definitions and summary statistics of the variables used in the analysis. For the dependent variable, we use the proportion of manure that is given away (y 1), sold (y 2) and self-used (y 3). In the survey, we collected the specific amount of manure that is produced and disposed of. The calculated portions of three disposal modes sum up to one. Of the three manure disposal modes, the average selling, giving way and self-use ratios are, respectively, 14, 32 and 55%.
Table 1. Variable definition and descriptive statistics.
The core explanatory variable in this study is the size of beef cattle farms. In the survey, we collected the beginning beef cattle inventory in 2018 of each farm as the measure of cattle farm size. This is consistent with previous studies that use the herd size or output volume as the measurement of farm size, e.g., the number of livestock and poultry actually raised at the end of every year (Yao and Zhang, 2021), the total livestock units the household owns (Araya, 2020) or the sum of the annual slaughter and the year-end inventory (Si et al., 2019a). The average farm size in our sample is 97.84 cattle. The appendix provides details on sub-categories of the variables based on the classification standard of China Animal Husbandry and Veterinary Yearbook (China's Ministry of Agriculture and Rural Affairs, 2021). For farm size, as shown in Table A1 of the appendix, nearly half of the observations are in the 10–49 heads category, about 22% are in the 50–99 heads category, 28% of the observations are large-scale farms (i.e., above 100 heads) and the remaining 2% are farms with 9 or fewer cattle.
In addition to the farm size variable, we also include a group of control variables based on previous literature. To reflect the influence of personal characteristics on behavioral decision-making, we include farmers' age, education and physical health status. To reflect the influence of family characteristics, we include proportion of family surplus labor, engagement in crop farming and acquisition of transferred land. To reflect the influence of farm characteristics, we include participation in agricultural cooperatives, livestock farming experience, the availability of manure treatment facilities, relevant training, the willingness of nearby farmers to accept cattle manure and the cognition on the income from manure recycling.
As shown in Table 1 and Table A1, the average age of farmers is 47.25 years, partly because cattle farming is labor demanding. The majority of farmers are in excellent (49.28%) or good (28.99%) health status. The proportion of household surplus labor in the total household population is 45% on average, mainly concentrated between 26 and 50%. The length of education averages 8.12 years, corresponding to finishing a junior high school. About 91% of the observations practice both crop and livestock farming, which could ensure the timely treatment or recycling of cattle manure. Cattle farmers on average have 7.01 years of farming experience, with about two-thirds having only 0–5 years of experience. For the availability of manure treatment facilities, 72% of the farmers in our sample have built or purchased manure treatment facilities, such as a fixed manure yard, an anti-seepage sewage pool, tertiary treatment facilities, organic fertilizer processing equipment or biogas digesters.
Results and discussion
Graphical analysis
Figure 2 displays a preliminary graphical analysis demonstrating how manure disposal modes change with farm size. For small farms, the proportion of cattle manure that is self-used is high, and that of giving away or selling is low. Especially for those farms with 9 or fewer cattle, manure is almost completely self-used, with a small amount given away to neighboring farmers. As farm size increases, the proportion of manure self-use shows a significant downward trend, from 86% at the size category of 9 or fewer cattle to 41% at the category of 100 or more cattle, a drop of over 50%. However, self-use is still dominant among the three disposal modes, even at the largest farm size. The average selling ratio of cattle manure shows a steady upward trend as farm size increases. The selling ratio of cattle manure at the category of 100 cattle and above is 30%, which is comparable to giving away in the same farm size category. The average giving away ratio increases and then decreases as farm size increases, but overall, the proportion of manure being given away remains at a relatively high level.
Fig. 2. Relationships between farm size and manure disposal modes.
Result analysis and interpretation
Table 2 reports the regression results of testing Hypothesis H1. In order to reduce the possibility of heteroscedasticity in the model, the logarithm is taken for the farm size variable. The Tobit model results are in regressions (1), (3) and (5), and the results of the constrained singular SUR model are in regressions (2), (4) and (6). Comparing the regression results of the Tobit and SUR models, the parameter value and significance level of explanatory variables are almost identical. In the results of the constrained singular SUR model, the P-value of the ‘no contemporaneous correlation’ test of the residual terms across equations is significant at the 1%, which justifies the use of SUR framework to improve the estimation efficiency.
Table 2. Results of the impact of farm size on the choice of manure disposal modes (I).
Notes: The values in brackets are standard errors; ***, ** and * indicate that explanatory variables are significant at the level of 1, 5 and 10% respectively. Since the results in regression (2) are calculated based on the results in regression (4) and regression (6) and the constraints between equations, the parameters other than independent variable coefficients and standard errors are not listed temporarily, considering the research needs of this study. If interested in the specific solution process, readers may contact the author.
As shown in Table 2, farm size is a significant determinant on farmers' choice of manure disposal modes except the giving away option. Farm size negatively impacts the percentage of self-use manure and positively impacts the percentage of selling. This indicates that the larger the farm size, the lower the proportion of manure self-use and the higher the proportion of selling through markets. This is consistent with the findings of Huang et al. (2016), Kong et al. (2016) and Pan et al. (2021), which supports Hypothesis H1. Larger farm size is associated with higher manure output, which exceeds the capacity of the farm-operated cropland to absorb it (Wang et al., 2021a, 2021c). Although some large-scale farms can rent some land for fodder production, they still lack sufficient land to absorb the large amount of manure generated and face severe challenges for the disconnection between cropland and livestock (Kuhn et al., 2020; Briukhanov et al., 2021; Pan et al., 2021). Therefore, the selling ratio increases with farm size. Meanwhile, the giving away ratio does not significantly change with farm size. This finding suggests that farmers do not resort to giving away as the solution for manure disposal, but primarily seek to maximize the economic return through selling manure.
Next, we test whether there is a nonlinear relationship between farm size and the choice of manure disposal modes. Based on Hypothesis H2, we introduce the quadratic term of farm size into the model specification and perform the regression analysis. Table 3 reports the corresponding results, where regressions (7), (9) and (11) are the results from the Tobit model, and regressions (8), (10) and (12) are the results of the constrained singular SUR model. Similar to the results in Table 2, we also find the residual terms across equations are correlated at the 1% significance level, which justifies the use of the SUR framework to improve the estimation efficiency.
Table 3. Results of the impact of farm size on the choice of manure disposal modes (II).
Notes: The values in brackets are standard errors; ***, ** and * indicate that explanatory variables are significant at the level of 1, 5 and 10% respectively.
The regression results in Table 3 identify the existence of a nonlinear relationship between farm size and manure disposal modes. For regression (8), the quadratic term of farm size is negatively significant at the 1% level, suggesting an inverted U-shaped relationship between farm size and the manure giving away ratio. For both the selling ratio in regressions (10) and the self-use ratio in regression (12), the quadratic term has a significantly positive impact on the dependent variable, indicating a U-shaped relationship. In other words, both the manure selling and self-use ratios first decrease and then increase with farm size, just contrary to that of the giving away ratio. In summary, a significant nonlinear relationship exists between farm size and farmers' choice of manure disposal modes, which supports Hypothesis H2.
The existence of this nonlinear relationship can probably be explained to a large extent by the difficulty of land acquisition or transfer due to China's land policy. All land in China is either state- or collectively owned, with urban land owned by the state and farmland collectively owned with a few exceptions (Ding, 2021). Before the Chinese government implemented a major land reform in its rural areas – the so-called ‘three rights (ownership, right to contract land and right to manage land) separation’ reform – farmers had concerns about land transfer due to unclear land rights and thereby were hesitated in participating in land transfer (Song et al., 2020; Fang and Guo, 2021; Yan et al., 2021). Although China's new land policy now eliminates farmers' concerns to lease land in order to promote large-size farms, there are still various barriers to farmland acquisition due to factors including information accessibility, policy cognition, application procedures and approval times. Farmland is an important carrier for manure disposal, and combined crop-livestock farming is recognized as a means of improving natural resource management and increasing agricultural productivity (Dos Reis et al., 2020; Ghimire et al., 2021; Wang et al., 2021a). Also, the scale of combined crop-livestock farming is an important factor affecting farmers' choice of manure disposal modes. As a cattle farm expands, manure production rapidly increases. With a fixed acreage of land for crop farming, the amount of manure exceeding the capacity of self-use also continues to increase, causing the proportion of manure self-use to decrease. With the further expansion of farm size, the potential economic value of manure increases due to its quantity. Moreover, the scale economies will be highlighted (Duvaleix-Tréguer and Gaigné, 2016; Hansen, 2019; Sharara et al., 2020). Therefore, farmers are incentivized to explore ways to maximize the economic return from manure disposal, such as leasing land to expand the scale of crop farming to increase self-use of manure as fertilizer or leasing land to install manure treatment facilities to promote profitable selling. This leads to the U-shaped relationship between farm size and manure selling/self-use.
For other control variables, the results also reveal several important findings. First, the educational level of a farmer has a positive and significant impact on manure selling, as in regression (10). This suggests that farmers with higher educational background can make better use of market to dispose manure and achieve higher economic values. More education can help farmers with better ability to obtain market information or develop sale channels. Second, the health condition of a farmer is positively related to the manure self-use ratio and negatively related to the proportion of given away. This is because the utilization of livestock manure is labor-intensive. Farmers that are in poor health condition may be less capable of disposing manure in a timely and effective way. To relieve the backlog of manure, it is necessary to give away to nearby farmers despite the lower economic gain. On the other hand, farmers in good health condition are more capable to dispose manure onto their own farmland. Third, engagement in crop farming is a significant determinant for manure disposal modes. As shown in regressions (8), (10) and (12), this variable is significant at 1% level. The effect is positive for self-use and negative for giving away and selling modes. This indicates that farmers give priority to applying manure to their own cropland as fertilizer. For the excess manure, they resort to external channels including selling and giving away. This suggests that an integration of crop and livestock farming can lead to more efficient disposal of manure. Fourth, land acquisition reduces the percentage of manure given away and increases the percentage of self-use. This is closely related to the previous finding that crop farming increases self-use of manure as fertilizer. Because acquired land is primarily used for crop production, land transfer could significantly increase manure disposal efficiency by promoting the self-use of manure to onsite cropland. Fifth, the availability of manure treatment facilities has a negative impact on manure self-use but shows no effect on manure giving away or selling. As shown in Table A1 of the appendix, although 71.74% of the beef cattle farmers own manure treatment facilities, most of the facilities are primary treatment facilities such as manure yards and sedimentation tanks, and only a few farmers are equipped with more advanced ones such as tertiary treatment facilities, biogas digesters and organic fertilizer processing devices. As a result, most farmers simply compost or dry manure to make it easy to transport. This may facilitate and encourage nearby farmers to accept manure. Therefore, the availability of manure treatment facilities tens to reduce manure self-use. Finally, the willingness of nearby crop growers to accept manure is an important factor affecting a farm's manure disposal mode. It significantly decreases self-use but increases the selling and giving away proportions. The willingness of crop growers to use cattle manure generates market demand and provides an opportunity to alleviate the burden of manure disposal on cattle farmers subject to environmental regulations.
Robustness test
As a robustness check of the empirical results, we use the MV Tobit model to re-estimate the regressions. Due to the add-up singularity condition of dependent variables, we conduct the MV Tobit model only on the two mostly used manure disposal modes, which are giving away (32%) and self-use (55%). We consider both the linear and quadratic specifications of the farm size variable. The regression results are summarized in Table 4. The MV Tobit (I) model only considers the linear relationship between the farm size and manure disposal modes, with the estimation results taken as a reference, while the MV Tobit (II) model introduces the square term of farm size. The correlation coefficients of residual terms in both the MV Tobit (I) and (II) models are significant at the 1% level, indicating that it is necessary to jointly estimate the equations to improve efficiency.
Table 4. Robustness test results.
Notes: The values in brackets are standard errors; ***, ** and * indicate that explanatory variables are significant at the level of 1, 5 and 10% respectively.
Comparing the results of the MV Tobit (II) model with that of Table 3, the parameter estimates are consistent in both values and significance levels. The farm size and its quadratic terms identify a significant inverted U-shaped relationship between farm size and the giving away ratio as well as a significant U-shaped relationship with the self-use ratio, which are consistent with the main empirical analysis results. Moreover, the estimated results of the control variables are consistent with those from regressions (8) and (12) in Table 3, demonstrating that the empirical results in this paper have good robustness.
Conclusions and implications
Based on primary survey data of beef cattle farmers in China, we use the constrained singular SUR model to empirically examine the effect of farm size on farmers' choice of alternative manure disposal modes, which include selling to neighboring farmers, giving away to neighboring farmers and self-use. The results demonstrate a significant and nonlinear impact of farm size on farmers' choice of manure disposal modes. Specifically, the percentage of manure for self-use or selling first decreases and then increases with farm size, indicating an inversed U-shaped relationship. On the other hand, the percentage of manure given away first increases and then decreases with farm size, indicating a U-shaped relationship. The results also reveal several other factors impacting farmers' choice of manure disposal modes, including the educational level and physical health condition of a farmer, engagement in crop farming, participation in land transfer, the availability of manure treatment facilities and the willingness of nearby farmers to accept manure.
We contribute to the literature in the following areas. First, existing studies on manure disposal modes mainly focus on the hog and poultry sectors, while few have studied the beef cattle sector. Because of the differences in the physical and chemical properties of animal manure across species, the disposal modes of cattle manure are different than those of hogs and poultry. Thus, this study on cattle manure disposal provides additional insights into current literature. Second, to identify the determinants of manure disposal modes, previous studies (He et al., 2016b; Li et al., 2021; Yao and Zhang, 2021) mainly rely on single-equation estimations such as Logit, Probit or structural equation methods. These methods cannot impose add-up constraints such as the sum of the three manure disposal ratios must equal to one. Additionally, the single-equation approach neglects the potential correlation of residuals across different equations, which reduces the estimation efficiency. To overcome these shortfalls, we use the constrained singular SUR method to jointly estimate the equations of manure disposal modes together and meet the add-up constraint. Third, despite farm size being recognized by previous literature as an important factor, none of the existing studies has considered its possible nonlinear relationship concerning manure disposal modes. The nonlinear relationship identified in this study helps explain the inconsistency in previous findings (Kong et al., 2016; Pan, 2016; Xu et al., 2016; Shu et al., 2017; Pan et al., 2021).
The findings of this study also have important policy implications for promoting efficient manure disposal in regions facing similar challenges of livestock manure management. First, there is no silver lining in improving livestock manure disposal across different farm sizes. Measures aimed at more efficient manure disposal should consider the heterogeneity in farmers' choice of disposal modes with regard to farming scale. For smaller farms, more manure for self-use or giving away should be encouraged to co-develop integrated crop-livestock systems with nearby farmers. For larger farms, incentives can be provided for investment in manure treatment facilities such as biogas digesters (Yang et al., 2019; Adghim et al., 2020; Sharara et al., 2020) and organic fertilizer processing equipment (Sharara et al., 2018; Makara et al., 2019; Zubair et al., 2020) to facilitate the storage, transportation and sales of manure. In addition, land transfer can be facilitated in rural areas to enable larger farms to lease in land to develop onsite integrated crop-livestock systems, which is consistent with the literature (Gowing et al., 2020; Ghimire et al., 2021; Wang et al., 2021a). Second, since larger farms are associated with manure disposal modes that generate higher economic return of manure utilization, expansion and consolidation of livestock farms may be encouraged through financial incentives (e.g., tax credits, low-interest loans and grants for facility infrastructure construction) and agricultural extension services (e.g., technical assistance and cooperatives). Third, integrated crop-livestock farming systems can be promoted through such means as facilitated land transfer. For regions that are intensive in livestock farming but lack land sale or lease markets, the development of such land markets should be supported. Lastly, given the willingness of nearby farmers to accept manure a significant factor impacting manure disposal modes, education programs can be designed and implemented to target plant growers (e.g., field crops, fruit trees and flowers) to advertise the benefits of using livestock manure as organic fertilizers.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1742170522000187
Acknowledgements
The contributions of Qian Li and Yubin Wang were based on the work supported by the National Social Science Foundation of China (No. 19CGL034) and the Capital Circulation Research Base Project (No. JD-KFKT-2021-004; JD-ZD-2021-002).
Conflict of interest
All authors declare that they have no conflict of interest.
