Shd Olid JIMOH , DING Wnqing, DONG Hibin, BAI Hihu,YIN Ynting, LIU Huihui, HOU Xingyng,

a Grassland Research Institute, Chinese Academy of Agricultural Sciences/Key Laboratory of Grassland Ecology and Restoration, Ministry of Agriculture, Hohhot, 010010, China

b Department of Pasture and Range Management, College of Animal Science and Livestock Production, Federal University of Agriculture, Abeokuta,23439, Nigeria

c Agriculture Research Group, Organization of African Academic Doctors (O-AAD), Nairobi, 25305-00100, Kenya

d The Party School of the Communist Party of China, Ningxia Hui Autonomous Region Party Committee, Ningxia Administration Institute,Yinchuan, 750021, China

e College of Pastoral Agricultural Science and Technology, Lanzhou University, Lanzhou, 730020, China

f Grassland College, Shanxi Agricultural University, Taigu, 030801, China

ABSTRACT

Recent researches have primarily focused on the relationship between livelihood strategies and livelihood capital, with few empirical studies on the sensitivity of livelihood strategies to livestock production and marketization in Inner Mongolia Autonomous Region of China. This study used an income distribution approach to categorize livelihood strategies of the respondents(n=394) into three types, i.e., herder livelihood strategy (LS1), petty-herder livelihood strategy (LS2), and non-herder livelihood strategy (LS3). Using the multinomial logistic regression model, we compared livestock production and marketization across the three livelihood strategies. Our findings showed that(1) livestock production and marketization tended to favor LS1; (2) an increase in the land asset (contracted and rented grassland) and off-take rate increased the probability of households choosing LS1; (3) stocking rate was higher for LS1; and (4) the higher critical market-related risks perceived by herders were animal price and hay and corn price. Moreover, higher livestock price acted as a deterrent to diversifying into other livelihood strategies (LS2 and LS3).Finally, this study advocates for policies that will promote the land transfer market, adopt modern techniques in animal husbandry, improve the medium for disseminating market information to herders, and provide incentives for long-term livelihood transformation.

ARTICLEINFO

Keywords：

Livelihood strategy

Livestock production

Marketization

Livelihood

diversification

Grassland

Inner Mongolia

Autonomous Region

1. Introduction

1.1. Concept of livelihood and livelihood strategy

Livelihood has been extensively discussed in the literature. The sustainable livelihood framework (SLF) defines livelihood as the combination of assets, activities, and the resources required by households to make a living (Ellis,1999, 2000; Su et al., 2018). Over the last few decades, international institutions (e.g., Department for International Development (DFID), Food and Agriculture Organization of the United Nations (FAO), etc.) have established similar conceptual frameworks (i.e., SLF) to study how factors such as assets, organizational structure, and institutional process influence livelihood (DFID, 1999; Fang et al., 2014). A rural livelihood is sustainable when it can meet its present needs without threatening the ability of future generations to do so (Scoones, 1998; Amadi and Anokwuru, 2017). However,while agricultural productivity is becoming increasingly dependent on climate, natural resource utilization has a significant impact on attaining rural livelihood goals (Ibrahim et al., 2015; Liu et al., 2018).

Livelihood strategy refers to a range of choices made by individuals or households (e.g., resource allocation and production activities) to generate income to pursue livelihood goals (Ellis, 2000; Nielsen et al., 2013; Fang et al., 2014;Liu et al., 2018). Households’ choice of livelihood strategy is determined by the type of assets owned (e.g., land and livestock) and how these assets are used to generate income (Su et al., 2009; Hua et al., 2017; Ding et al., 2018).Consequently, herders’ choice of livelihood strategy impacts the welfare, safety, and individuals’ ability to cope with income shocks (Rakodi, 1999; van den Berg, 2010). Several studies on the nexus between livelihood strategy and rural development have emphasized the significance of livelihood assets by focusing on how the latter influences herders’choice of livelihood strategy (Ellis, 2000; Zhu et al., 2013; Majekodunmi et al., 2017; Su et al., 2018). More importantly,household enterprise productivity and marketization of outputs are critical to maintaining, diversifying, or settling for an alternative livelihood strategy. However, only a few quantitative studies have attempted to investigate this topic in the context of household units in Inner Mongolia Autonomous Region, China.

1.2. Factors affecting livelihood strategy

The conceptual divergence in the classification of livelihood strategy has led to significant progress in understanding various livelihood patterns in different environments, resulting in poverty reduction interventions (Fang et al., 2014;Sun et al., 2019; Huq et al., 2020). As a result, livelihood analysis has taken center stage in rural development studies,with a growing recognition that assets (i.e., livelihood capitals) govern livelihood diversification or transition by households (Ellis, 2000; Scoones, 2009; Nielsen et al., 2013; Peng et al., 2017), allowing resilience to shocks and stresses (Rakodi, 1999). Recent academic studies, on the other hand, have increasingly highlighted the factors influencing livelihood strategy. These factors included natural and human capital (Hua et al., 2017; Liu et al., 2018),financial and social capital (Peng et al., 2017; Su et al., 2018), ecosystem-based activities (Huq et al., 2020), policy(Ellis, 1999; Kassie et al., 2017), and marketization and income streams (Yang et al., 2018; Sun et al., 2019). Besides,Nielsen et al. (2013) used the activity choice approach to quantify rural livelihood strategy in three developing countries,i.e., Nepal, Bolivia, and Mozambique. They found a similar level of reliance on the environment across all livelihood strategies. They also suggested that, while non-farm income correlates with higher earnings, a higher level of specialization does not guarantee a more lucrative livelihood strategy.

Moreover, our literature review showed that using a quantitative approach to study livelihood diversification and its influencing factors has become popular in academia. Kassie et al. (2017) investigated the factors that compel farmers to participate in non-agricultural income diversification activities in Ethiopia, and concluded that institutional factors,such as land tenure security and membership in cooperatives, are important for livelihood diversification. In Nigeria,the impact of education and multiple sources of income on livelihood strategy has been documented (Adepoju and Obayelu, 2013; Oyinbo and Olaleye, 2016). According to Gebru et al. (2018), diversification into non-farm strategies is critical to rain-fed agriculture, the dominant mode of production in sub-Saharan Africa; the authors concluded that government policies designed to support livelihood diversification concerning national job creation are critical.Similarly, the importance of livelihood capitals on livelihood strategy has been widely emphasized in China, and this topic has sufficiently improved global understanding of the relationship between livelihood strategy and sustainable development in the rural areas (Fang et al., 2014; Liu et al., 2018; Su et al., 2018; Yang et al., 2018). However, it is worth noting that few studies have examined these relationships concerning herders (e.g., Achiba, 2018; Ding et al.,2018), implying the need for more livelihood studies about this group of people.

1.3. Livestock production and marketization in Inner Mongolia

In China, grasslands in Inner Mongolia represent one of the epicenters of grazing-based livestock production (Hou et al., 2014; Hu et al., 2019). From 1979 to 2013, the region accounted for 6% of China’s cattle production and 8% of total beef production, ranking the second in both cases (Liu et al., 2017). Also, it is one of the significant mutton producing regions in China. As pointed out by Liu et al. (2017), Inner Mongolia ranked the first in 2013, accounting for 18% of the Chinese sheep population and 22% of total mutton output in the country. In addition, mutton output in 2013 was 14 times that in 1979. However, overgrazing was widely perceived responsible for the decline of grassland productivity and carrying capacity in Inner Mongolia by 40% and 60% in the 2010s, respectively, compared to the 1950s (Wang, 2007). Briske et al. (2015) discovered, for example, that grasslands only provide 24% of the fodder requirement of livestock for bodyweight maintenance and that stocking rate exceeds theoretical carrying capacity by 3.2 times, resulting in land degradation. This evidence supported the finding of Hu et al. (2019) that grassland degradation in Inner Mongolia has reduced livestock productivity. Further, the introduction of intensive livestock production in Henan and Shandong provinces of China is threatening the competitiveness of animal production in Inner Mongolia of China (Liu et al., 2017). Diversification into non-herding livelihood strategy has become popular in this region (Squires et al., 2009), including engaging in business activities (Walelign, 2016; Ding et al., 2018), developing ecotourism (Zhang et al., 2019), and working in city-based companies. These changes in livelihood have an impact on livestock husbandry development in the region.

Over the past three decades, Inner Mongolia has developed market-supporting institutions aimed at long-term sustainable marketization of livestock to 1.3×109consumers (Lohmar et al., 2009; Byne, 2016; Li et al., 2017). This marketing system is efficient at the expense of substantial incentives for producers whose production decisions are based on their knowledge of future market trends (Lohmar et al., 2009; Roessali et al., 2011). Consequently, herders’livestock production practices (e.g., lambing time) and marketing strategies (e.g., selling animals in the same year of lambing) have changed dramatically (Li et al., 2017). Academic studies have shown that grassland productivity, animal market prices, geographical location, and livestock holding structure all impact pastoral livestock production and marketization (Du et al., 2017; Li et al., 2017; Ding et al., 2018; Hu et al., 2019). For example, in response to lower selling prices, herders will rationally refuse to sell livestock, resulting in higher feed and health-related costs and lower profit margin when the animals are eventually sold (Byne, 2016). Therefore, the shift in livestock production from Chinese pastoral areas to crop-livestock farming areas, the reduction in livestock market share from grazing systems,and other unfavorable market forces (e.g., market price) could potentially influence herders to diversify their living strategy to secure a sustainable livelihood (Li et al., 2008; Squires et al., 2009; Zhu et al., 2013; Wang et al., 2016).

1.4. Justification and significance of the study

In recent decades, Inner Mongolia grasslands have experienced changes in climate events, which altered the climatepasture-livestock system (Piao et al., 2010; Hou et al., 2012; Li et al., 2017). Aside from human activities, climate change has led to a local mismatch between livestock requirements and forage supply by grasslands, resulting in overgrazing. Therefore, Inner Mongolia is significant for studying the relationship of livelihood strategy with livestock production and marketization owing to its vast grassland area, dense livestock and human population, severe ecological degradation, and its contribution to the environmental security of Chinese grasslands (Liu et al., 2017; Su et al., 2018).

There is a growing interest in quantifying the relationship between livelihood strategy and livelihood capitals as a means of developing poverty reduction policies that promote sustainable livelihood in developing countries (Ellis, 2000; Kassie et al., 2017; Yang et al., 2018). In this sense, many studies have identified how the five livelihood capitals (i.e., human,natural, physical, financial, and social capitals) affect herders’ choice of livelihood strategy (Nielsen et al., 2013; Fang et al., 2014; Liu et al., 2018). There is, however, a scarcity of empirical data on the sensitivity of livelihood strategy to livestock production and marketization. Notably, there are few analytical studies on this topic in Inner Mongolia.

Agriculture employs 60% of rural dwellers in the Chinese pastoral areas (including Inner Mongolia) (Zhou and Zhao,2020). Nonetheless, non-agricultural activity is gaining traction as a platform for poverty reduction due to the reform and opening-up policy in China (National Bureau of Statistics of China, 2011). According to van den Berg (2010),starting with a profitable livelihood strategy does not guarantee the livelihood strategy because households can actively switch between livelihood strategies based on prevailing climatic, ecological, and economic conditions. These changes are critical for environmental security, sustainable livelihood, and rural development (Liu et al., 2018). Thus, this paper adds to our understanding of livelihood strategy by providing a detailed statistical analysis of various livestock production and marketization variables and, more specifically, investigating the sensitivity of livelihood strategy to these variables in Inner Mongolia. We are explicitly interested in understanding the role of livestock production and marketization in the choice of livelihood strategy for pastoralist families. More specifically, the paper seeks to (1) assess the sensitivity of livelihood strategy to livestock production and marketization, in contrast to previous studies that focused on the relationship between livelihood strategy and livelihood capitals; (2) employ an apt econometric model(i.e., multinomial logistic regression model) to reveal the factors affecting livelihood strategy at the household level;(3) evaluate herders’ perceptions of market-related risks; and (4) discuss the policy implications of the findings. This is significant from the standpoint of socio-ecological systems (i.e., coupled human-natural environment system) (Ojima et al., 2013; Huntsinger and Oviedo, 2014), which are critical to herders’ livelihoods and sustainable utilization of grasslands in Inner Mongolia (Zhang et al., 2019). The findings are applicable to Inner Mongolia and have significant implications for the sustainability of grasslands in the arid and semi-arid regions worldwide.

2. Methodology

2.1. Study area

Inner Mongolia Autonomous Region is located in northern China, which has an arid and semi-arid climate and an average elevation of 1000 m. The region has 118.30×104km2of land, representing 12.3% of China’s total land area(Ding et al., 2018). Grasslands in this region cover 7.90×104km2(Xue et al., 2020), accounting for 60.0% of the land area in Inner Mongolia and 21.7% of permanent grassland area in China; they provide critical ecological services in northern China (Liu et al., 2017; Ding et al., 2018; Zhang and Brown, 2018). Given the extensive grasslands, the local households mainly raise sheep, cattle, and goats, which play an essential role in supplying animals and animal byproducts throughout the region and China. However, some households combine intensive livestock husbandry with cropping, a practice known as mixed-crop livestock production (Waldron et al., 2010).

2.2. Methods

2.2.1. Sampling distribution and data collection

We selected study participants using a stratified random sampling procedure during the household survey in the meadow steppe, typical steppe, and desert steppe in Inner Mongolia. For each grassland type, two counties were selected. At the county level, 2-3 townships were chosen, and 2-3 villages were selected in each township, with at least 60 households sampled in each village in proportion to the total number of households. We used a semi-structured questionnaire in conjunction with participatory rural appraisal to collect data from the selected households during interviews from September 2018 to December 2018. The survey questionnaire focused on the socio-economic information of households, livestock holding structure and grassland areas (contracted and rented), livestock production activities (e.g., lambing rate), and marketization (e.g., sheep price). The variables are described in detail in Table S1.We used simple language and short sentences to ensure clarity and ease of administration during questionnaire development. A total of 450 questionnaires were collected, of which 433 completed responses can be used in this study,with a response rate of 96.2%. Further, we excluded some questionnaires with missing observations and quality issues,leaving the final 394 valid questionnaires for further analysis.

2.2.2. Descriptive analysis

The respondents’ socio-economic characteristics were reported using descriptive analysis by percentages. We used a two-step procedure to determine herders’ interpretation of the potential impact of market risk (i.e., risk perception)(Pennings et al., 2002). First, a set of market-related risks was read to the respondents in perceptual statements to learn about their opinions on each variable. These variables included animal price, hay and corn price, information on animal price, no precise information on animal price, and other living costs. Second, households were asked to rate the severity of each variable on a five-point scale. A score of 1 indicated the most severe, while a score of 5 implied the least severe.This method was consistent with Ibrahim et al. (2015).

2.2.3. Livelihood strategy classification

There is a growing interest in the study of livelihood strategy across the globe, and livelihood classification has been explored extensively (Zhang et al., 2008; Nielsen et al., 2013; Zhang et al., 2013; Fang et al., 2014; Liu et al., 2018;Huq et al., 2020). Some of the methods in use included income-based approach (Ding et al., 2018; Yang et al., 2018),activity choice approach (Zhu et al., 2013; Peng et al., 2017; Sun et al., 2019), and asset-based approach (Fang et al.,2014; Hua et al., 2017). However, classifying farmers or households into livelihood strategies based on income structure is common in China (Ding et al., 2018; Liu et al., 2018). Pioneered by the Chinese Academy of Social Sciences in 2002, this method classifies households that derive 95.0% of their total income from agriculture as farm households,those that derive 95.0% of their total income from non-agricultural sources as non-farm households, and those that derive 5.0%-95.0% of their total income from non-agricultural sources as part-time households (Liu et al., 2018). The National Bureau of Statistics of China modified the method in 2004 by lowering the income threshold for farm households from 95.0% to 90.0% (Hua, 2014).

Relative to the unstable structure of the livestock market in the pastoral areas in recent years, herders’ income has become highly stochastic (Ding et al., 2018). Based on the summary of previous findings and the current situation in our study area, we used the percentage of livestock income (e.g., sales of animals and animal by-products, such as wool and cashmere) in herders’ total income to classify livelihood strategy in this study. On this basis, we divided livelihood strategy of herders into three types according to the method proposed by Yang et al. (2018). Specifically, herder livelihood strategy (LS1) was defined as households deriving ＞75.0% of their income from livestock, petty-herder livelihood strategy (LS2) was defined as households deriving 25.0%-75.0% of their income from livestock, and nonherder livelihood strategy (LS3) was defined as households deriving ＞75.0% of their income from non-livestock activities. This was consistent with the classification of Ding et al. (2018).

2.2.4. Data standardization

Before analysis, standardizing data can eliminate dimensional relationships between variables (Liu et al., 2018). This helps to compare data on a scale of 0-1 while measuring the relative contribution of each variable during analysis (Fang et al., 2014). Therefore, we normalized the continuous variables using the dispersion normalization described by Hua et al. (2017) and Liu et al. (2018), as shown in Equation 1:

where x means the variable (e.g., stocking rate) after standardization; X represents the variable before standardization;Xmaxis the maximum value of the variable; and Xminrefers to the minimum value of the variable.

2.2.5. Sensitivity of livelihood strategy: application of multinomial logistic regression model

In this study, livelihood strategy was defined as a combination of activities that generate income for households to sustain their livelihood (Ellis, 1999). Therefore, the causal relationships between the designated livelihood strategy and livestock production and between the designated livelihood strategy and marketization were modeled using multinomial logistic regression model in SPSS 19.0. The logit model was widely employed as a simple model to identify the determinants of livelihood strategy, where the dependent variable was categorical (Fang et al., 2014; Hua et al., 2017; Sun et al., 2019). The independent variables were livestock production and marketization, while the dependent variable was herders’ livelihood strategy (LS1, LS2, and LS3). We selected LS1 as the reference strategy for discussing the barriers of livelihood transition to LS2 or LS3. The logit model simulated herders’ selection of a livelihood strategy based on benefit maximization. The probability of choosing one of the three livelihood strategies is:

where ρ is the herders’ selection of a livelihood strategy; e is the base of natural logarithms; Bjmis a vector of coefficients; and Xmis the vector of explanatory variables (Field, 2009; Walelign, 2016). Given that herders have the options of LS1, LS2, and LS3, the expression for the latent variable function that herders choose from j=1, 2, 3 is:

In evaluating the model with LS1 as the reference, the obtainable logit formulas are:

wherepy1represents LS1;py2represents LS2;py3represents LS3; andx1,x2, …,xirepresent the explanatory variables.In Equations 3-5, ifj=2, thenαj=b210; ifj=3, thenαj=b310.b210andb310are constant terms; andb211,b212, …,b21mandb311,b312, … ,b31mare the estimated coefficients. We used the estimated coefficients to interpret the changes in the dependent variable caused by a unit change in the independent variables. When the coefficient of each independent variable is greater than zero while holding other variables constant, the variable(s) will influence herders’ choice of livelihood strategy. We also defined the change in odds due to a unit change in the independent variables as the sensitivity of livelihood strategy to livestock production and marketization variables (Field, 2009; Fang et al., 2014).The values greater than 1 indicated that as the predictor increased by one unit, the odds of an outcome increased, andvice versa(Field, 2009). Finally, we constructed two logistic regression models totally (each one for livestock production and marketization).

2.2.6. Test for multicollinearity

Multicollinearity denotes a perfect linear relationship between the independent variables, leading to redundant information in regression models (Landau and Everitt, 2004). To overcome this problem, we examined the independent variables’ tolerance and variance inflation factor (VIF). Tolerance was less than 1.00 and the VIF was less than 5.00 for both livestock production and marketization variables (Tables S2 and S3), indicating no multicollinearity between the independent variables, with no effect on the model analysis.

2.2.7. Estimation of lambing rate and selling rate

The lambing rate, calving rate, and livestock selling rate for both livestock species (i.e., sheep and cattle) considered in this study were estimated using the following equations proposed by Li et al. (2017):

where LR represents lambing rate (calving rate for cattle);Nlambis the number of lambs produced in the year under consideration (calves for cattle); andNeweis the number of ewes stocked in the current year (cows for cattle).

where SR is the selling rate;Nsoldis the number of livestock sold; andNtotalis the total number of livestock in the year under consideration. We applied the equation to each livestock species separately.

3. Results

3.1. Summary statistics of surveyed households

The result of the household interviews showed that the study area tended to be homogeneous in terms of household head’s age, ethnicity, and educational level, as well as the households’ distance to the road and city, with minor variations across different grassland types. The proportions of male household heads were 80.0%, 73.7%, and 76.1% in the meadow steppe, typical steppe, and desert steppe, respectively (Table 1). The average age of the household heads ranged from 46.0 to 47.0 years across the grassland types. Households in the meadow steppe lived closer to the road(14.2 m) and city (40.7 km) than those in the typical steppe (25.6 m and 48.1 km, respectively) and desert steppe (20.2 m and 48.0 km, respectively). More households worked in villages in the typical steppe (35.8%) than in the desert steppe (14.1%). The Mongolian ethnic group accounted for 88.0%, 78.8%, and 85.9% of the surveyed households in the meadow, typical, and desert steppes, respectively. On average, a larger percentage of the respondents in the meadow steppe attended middle school (40.8%), and those in the typical and desert steppe attended high school (40.1%).

Table 1Summary statistics of the surveyed households.

We described the model results using the regression coefficients and odds ratio (Exp (B)) values for each variable.The analysis indicated that there were remarkable relationships between livelihood strategy (i.e., dependent variable)and livestock production (i.e., independent variables) and between livelihood strategy and marketization (i.e.,independent variables), with the likelihood ratio chi square values of 52.31 (df=10;P＜0.000) and 52.98 (df=20;P＜0.000) for model 1 (Table 2) and model 2 (Table 3), respectively. This demonstrated the high goodness of fit for the two models, implying that the models were broadly consistent and the estimated variables were stable and credible.

Table 2Results of the relationship between livelihood strategy and livestock production based on the multinomial logistic regression model 1.

Table 3Results of the relationship between livelihood strategy and livestock marketization based on the multinomial logistic regression model 2.

3.2. Sensitivity of livelihood strategy to livestock production and marketization

3.2.1. Sensitivity of livelihood strategy to livestock production

As shown in Table 2, an increase in lambing rate was required for households to adopt LS2. This implied that every unit increase in lambing rate would increase the likelihood of switching to LS2 by 4.023. Similarly, a decrease in contracted grassland area increased the chances of adopting LS3 by 0.024. We found that rented grassland area and stocking rate significantly influenced LS1 during the transformation of LS1 to LS2 and LS3, respectively (Fig. 1). This finding indicated that reducing the rented grassland area by one unit could increase the likelihood of choosing LS2 and LS3 by 0.090 and 0.005, respectively. Furthermore, if a household’s stocking rate decreased by one unit, the chances of adopting LS2 and LS3 increased by 0.065 and ＜0.001, respectively, indicating that a high stocking rate was a barrier for households to choose LS2 and LS3.

Fig. 1. Schematization of the relationship of livelihood strategy with livestock production and marketization. The variables in solid fill grey boxes are those (e.g., lambing rate and sheep selling rate) that influenced the livelihood strategy of the surveyed households.

3.2.2. Sensitivity of livelihood strategy to livestock marketization

Here, we analyzed the sensitivity of livelihood strategy to livestock marketization across the study area. Interestingly,LS3 had a lower sheep selling rate, cattle selling rate, and sheep price compared to LS1. Regarding the odds ratio value,households may prefer to pursue LS3 given the low sheep selling rate and cattle selling rate. A unit increase in sheep price could reduce the likelihood of shifting from LS1 to LS3. Compared with LS1, the only market-based variable that compels households to adopt LS2 was lower cattle price. A unit increase in cattle price reduced the chances of households to choose LS2 over LS1.

3.3. Herders’ perception of market-related risk

The combination of our analyses and field observations revealed that herders’ perceptions of market-related risks were different. This could be a result of the diverse local environmental and market situations across the grassland types. About 60.8%, 91.2%, and 92.3% of the respondents ranked animal price as the primary market-related risk in the meadow, typical, and desert steppes, respectively (Table 4). Most of the respondents ranked hay and corn price as the second market risk. Only a small percentage of the respondents thought hay and corn price were the most severe market risk (9.5% in the meadow steppe, 4.4% in the typical steppe, and 6.3% in the desert steppe). Comparatively,whether herders receive information on animal price or no precise information on animal price was perceived as a less critical market risk by the respondents. Moreover, few of the respondents in the meadow steppe (21.6%) and a very small portion of the respondents in the typical steppe (2.2%) and desert steppe (0.7%) believed that other living costs significantly contributed to the market-based challenges confronting herders. However, 16.8%, 14.6%, and 45.1% of the respondents agreed that other living costs were an intermediate risk that could affect their livestock marketability.

Table S1Livestock production and marketization variables used in multinomial logistic regression model.

Table S2Tolerance and variance inflation factor (VIF) of livestock production variables.

Table S3Tolerance and VIF of livestock marketization variables.

Table 4Herders’ perception of market-related risk across the grassland types.

4. Discussion

We used the proportion of livestock income in total household income to classify households’ livelihood strategy into herder livelihood strategy (LS1), petty-herder livelihood strategy (LS2), and non-herder livelihood strategy (LS3)to estimate the sensitivity of livelihood strategy to livestock production and marketization. Most of the households in Inner Mongolia relied on livestock production (Li et al., 2017; Liu et al., 2017). This reliance is prone to risk depending on livelihood strategy, livestock productivity, and marketability potential. However, few researchers have studied such relationships in Inner Mongolia, the major livestock production base in China (Hou et al., 2014; Liu et al., 2017). Our results contribute to the literature on the relationship of livelihood strategy with livestock production and livestock marketization. Therefore, the results of this study were anticipated to provide some information for livestock production management and livestock marketing-related policymaking. Our research focused on the meadow steppe, typical steppe, and desert steppe, however, a broader study area that includes other grassland types in Inner Mongolia and other grassland areas in northern China could be incorporated in future studies.

We presented three significant findings that may be potentially useful for future studies on livelihood diversification options to foster a sustainable livelihood in the pastoral areas of Inner Mongolia and other regions with similar ecosystems. Our research findings can be summarized as follows: (1) LS1 represented an integrated coupling of high rented grassland area and high stocking rate on grasslands; (2) households with a low contracted grassland area adopted LS3, indicating a shift away from intensive livestock production; and (3) better livestock marketization appeared to favor the adoption of LS1 by households relative to off-take rate (sheep selling rate and cattle selling rate) and animal price (sheep and cattle price), compared to LS2 and LS3. The relevance of these findings could be extended within the context of socio-ecological sustainability given the increasing differentiation of LS1 in Inner Mongolia and how this could impart natural resource utilization (Peng et al., 2017; Ding et al., 2018; Liu et al., 2018; Huq et al., 2020). As Zhang et al. (2020) proposed, the concept of ‘Sustainability of the Grassland’ promotes the integration of the soil,grassland, animal, and human as a coupled socio-ecological system, which is critical to sustainable natural resource utilization.

4.1. Impacts of livestock production on herders’ choice of livelihood strategy

A good understanding of livestock production and management (i.e., household livelihood strategy) was required for adopting LS1, which was reflected in variables such as lambing rate and calving rate. We found that an increase in lambing rate can promote the likelihood of households to choose LS2. This demonstrated that a focus on lambing rate was a critical indicator for improving LS2 (Fang et al., 2014). Furthermore, Li et al. (2017) reported that increased lambing rate was positively correlated with a shorter turnaround time for livestock production due to the regulation of the mating period. This signified the importance of familiarity with essential management practices in livestock production for sustainable livelihood and development (Fang et al., 2014; Yang et al., 2018; Sun et al., 2019). The contracted grassland area did not change between LS1 and LS2, suggesting that contracted land asset was critical for households engaged in those livelihood strategies (Nielsen et al., 2013). Lower contracted grassland area propelled herders to shift from LS1 to LS3. Numerous studies have reported the significance of contracted grassland areas for households engaged in livestock production to seek a sustainable livelihood (Fang et al., 2014; Peng et al., 2017; Ding et al., 2018; Liu et al., 2018; Yang et al., 2018).

The sustainability of livestock grazing on Inner Mongolian grasslands has been widely debated. There was no consensus between the government recommended stocking rate and the actual stocking rate used by herders (Zhu et al., 2013; Hou et al., 2014; Li et al., 2017). Our findings suggested that for LS1, households need to rent-in larger grassland areas and use a high stocking rate. This result appeared intriguing and contrasting, because one would expect increased stocking rate in grazing systems to be buffered by access to additional land area. However, whether the anticipated equilibrium state was feasible and sustainable from an ecological perspective merits further research. Other studies have shown that households who did not engage in intensive livestock production grazed less (Majekodunmi et al., 2016; Hua et al., 2017; Hu et al., 2019). More importantly, promoting a land circulation system could help households with LS1 in increasing their access to land resources (Zhu et al., 2013; Hu et al., 2019), practicing modern livestock production (Lohmar et al., 2009; Ibrahim et al., 2015), reducing grazing pressure (Hou et al., 2014; Yin et al.,2019), and improving the sustainability of household and grassland ecosystem.

4.2. Impacts of livestock marketization on herders’ choice of livelihood strategy

Marketization is critical for households’ long-term adaptation to socio-ecological system fluctuations (Li et al., 2017).This has led to the independent and sedentary mode of livestock production in our study area (Li and Huntsinger, 2011).Thus, our analyses of the sensitivity of livelihood strategy to livestock marketization can advance the understanding of the relationship between the former and the latter and provide insight into market-based policies to improve herders’livelihoods. The off-take rate (sheep selling rate and cattle selling rate) was important for livelihood differentiation from LS1 to LS3 but not for livelihood differentiation from LS1 to LS2. Similar findings have been reported in Nigeria(Majekodunmi et al., 2017) and China (Fang et al., 2014; Li et al., 2017). Therefore, promoting LS1 will necessitate continuous sensitization of herding households on modern livestock management practices. Practically, this implied putting a greater emphasis on livestock holding structure, livestock breeding (e.g., mating period), and lambing practices for increased productivity (e.g., shortened duration of livestock growth and selling of sheep in the same year)(Li et al., 2017; Yin et al., 2018). This also had implications for the sustainable development of the pastoral areas.

Our findings also showed that households made livelihood decisions based on animal price. Other researchers have made comparable quantitative observations in Kenya (Onyango, 2017) and western mountainous areas (Liu et al., 2018)and Inner Mongolia (Zhu et al., 2013; Hu et al., 2019) in China. More specifically, we discovered that herders who sold their sheep and cattle at a higher price were more likely to choose LS1, whereas those who responded to lower selling prices (for both sheep and cattle) were more likely to select LS2 and LS3. A plausible explanation for this was that animal price was critical to the sustainability of households with LS1, whereas households with LS2 and LS3 may engage in other non-herding activities to supplement their income and compensate for the lower selling price (Zhu et al., 2013; Byne, 2016; Liu et al., 2017). Although animal price influenced herders’ livestock supply decisions (Komarek et al., 2012; Liu, 2017), there was a disconnect between households and the market, severely limiting the effectiveness of policies to improve rural livelihoods (Brown et al., 2009). More importantly, local government policies targeted at the animal husbandry market were limited (Miao et al., 2018). As a result, Inner Mongolia required forage-livestock policies to reduce production risks, promote sustainable grassland utilization, and improve herders’ livelihood by connecting households and the market (Brown et al., 2009).

4.3. How do herders perceive market risk?

Market-based risks were associated with price changes (Ellis, 1998; Korir, 2011; The World Bank, 2016), and herders’ participation in the livestock market was critical for the sustainable economic development in pastoral areas(Flaten et al., 2005; Legesse and Drake, 2005; Boughton et al., 2007). The most critical market-based risk perceived by herders across the grassland types was animal price. This was because herders always seek higher prices per sale to meet their cash needs and sustain their livelihood (Adriansen, 2006; Majekodunmi et al., 2017). Huong and Nanseki(2015) and Bishu et al. (2018) also reported that animal price was an essential risk source for livestock households in Vietnam and northern Ethiopia. Other living costs (e.g., food, clothing, and energy) were also perceived as a significant risk source for households, consistent with the finding of Wang et al. (2013). The perception of hay and corn price risk was higher in the typical steppe and desert steppe than the meadow steppe. This was due to the meadow steppe’s superior grassland, which can produce more hay than other steppe types. Thus, hay price was likely to be higher in the typical and desert steppes. Bishu et al. (2018) also reported that fodder price was a significant concern for livestock producers in northern Ethiopia. As pointed out by Miao et al. (2018), the local government of Jalaid Banner in Inner Mongolia demonstrated a willingness to regulate animal market prices (e.g., price regulation and market trade control).Such regulation was needed across the region’s local governments to reduce herders’ production risk and increase profit margins. This suggests that more research is needed to determine how local government market-oriented policies affect herders’ livelihoods, livestock production risks, and marketization in Inner Mongolia.

Whether herders receive information on animal price or do not receive precise information was a lesser risk to households. Nevertheless, herders in the livestock business could benefit from adequate regulatory and animal price information. In this regard, the government (particularly at the local level) should strengthen the transmission of marketrelated information to herders through extension agents, herder groups, technologies (e.g., mobile smartphones), and improved infrastructures (Huong and Nanseki, 2015; Ibrahim et al., 2015; Su et al., 2018; Yang et al., 2018).

4.4. Policy implications

Based on our findings, we make some policy recommendations. To begin, it is worth noting that livelihood differentiation from LSI to LS2 and LS3 could potentially enhance the attainment of sustainable livelihood and land use in the study area (Zhang et al., 2008; Ding et al., 2018). However, scholars have raised concerns about promoting or restricting this form of differentiation (Hua et al., 2017; Liu et al., 2018). We suggest that government policies should aim to address the inherent needs of each type of households. For instance, households are more likely to become herding households on a larger scale, but they require more land resources to lower their stocking rate for environmental sustainability. The government should promote the land transfer market to aid the expansion of livestock production to ease adopting modern animal husbandry techniques and increase economic benefits to households. There has been a significant improvement in livestock marketization across Inner Mongolia (Li and Huntsinger, 2011; Li et al., 2017);this trend needs to be strengthened with policies focusing on better livestock market structures (e.g., higher animal price and lower fodder cost) and the adoption of various means of disseminating market-related information to households.There is an opportunity to capitalize on livestock markets in China’s eastern and southern cities, but it requires road infrastructure development to connect Inner Mongolia to these markets (Briske et al., 2015).

Government policies should focus on making incentives for livelihood transformation available to LS2 and LS3 households with lower land area and share of the livestock market (e.g., improving their education and skill acquisition).This would assist them in maintaining their livelihood after the transition. However, the promotion of sustainable livelihood at the regional and national levels depends on livelihood diversification (Zhang et al., 2008, 2013; Ding et al., 2018; Hu et al., 2019); policymakers believe that livestock development is a critical mechanism in achieving sustainable regional development (Komarek et al., 2012).

5. Conclusions

This study used a multinomial logistic regression model to quantify the sensitivity of livelihood strategy to livestock production and marketization in Inner Mongolia, where livelihood diversification and transition are common. Our findings showed that: (1) livestock production and marketization differed for different types of livelihood strategy, and tended to favor LS1; (2) the sensitivity of LS2 to livestock production was 4.023 for lambing rate, 0.090 for rented grassland area, and 0.065 for stocking rate; in contrast, it was 0.024 for contracted grassland area, 0.005 for rented grassland area, and ＜0.001 for stocking rate for LS3; and (3) low sheep and cattle selling rates propelled households to pursue LS3. We found that increased land asset (contracted and rented grassland) and livestock marketization increased the likelihood of households to choose LS1. These findings provide insight into how livestock production and marketization influence the types of livelihood strategy adopted by households and serve as a foundation for future research in this area. Our results shed light on the sustainable rural pastoral development in Inner Mongolia of China and other developing countries where similar livelihood strategies are pursued as a part of development policy. Finally,the government should encourage livelihood differentiation through education and skill acquisition incentives to smoothen diversification or transition, sustain the diversified or transiting households, and promote environmental sustainability.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research was supported by the Graduate School of Chinese Academy of Agricultural Sciences Scholarship(2017Y90100124), the National Natural Science Foundation of China (71774162), and the National Key Basic Research Program of China (2014CB138806).