4.1. Farm Household Income Generation Activities
In this sub-section, the composition of household income is studied. To identify main income generating activities, we classify rural households’ income (
Table 2) according to which sources they derive income from—farm or non-farm activities based on field observation in Xinghe. The main farm incomes include crop income (C), potato income (P), vegetable income (V), and livestock income (L); non-farm incomes including non-farm wage income (N) and state grants (S).
Crop income is the most basic farm income. Crop production provides a basic food source, and surplus crop can be exchanged for cash to improve living conditions. According to the interviews, the main crops in Xinghe are spring wheat, naked oat, benne, millet, and corn. The growing season of these crops is from May to September, but dry weather in this period often leads to low yield; for instance, the yield of benne is 300 kg/ha with low rainfall, while the yield reaches 700 kg/ha with normal rainfall. However, each family has about 2.5 ha of farmland on average. In other words, the advantage of the land size can compensate for the lower yield. The survey shows that crop sale could bring in about RMB 3000 (about 500 US dollars).
Potato income is another important source of farm income. Thanks to the soil and climate conditions in Xinghe, potato has excellent yield (about 12,000 kg/ha). Farmers could sell surplus potatoes for cash. In local markets, the average price of potato is RMB 1 per kilo, and most households could sell 3000–4000 kg during the harvest season and next spring.
Vegetable income is also an important source of income for mid plain areas. According to field observations in TJ, the income from cabbage and sugar beet could account for more than 30% of household income.
Livestock farming, mainly cattle and sheep or goats, is an important farm activity in northern hilly areas and southern mountain areas. According to the surveys in DZ and SWS, where seasonal livestock trade fairs are held each year, a head of cattle can be sold for RMB 4000 in DZ, and a sheep can be sold for RMB 200 in SWS.
Non-farm wage is a significant income source for households with young members. As wage labor, they enjoy stable monthly income of about RMB 2000 or more. Some local respondents said that they or their children could contribute about RMB 3000–4000 every year.
State grants are the least important source of income in Xinghe. In China, the elderly and disabled generally receive state pensions. In 2009, the monthly state pension per person was RMB 100; in 2013, it was RMB 150; now, it is RMB 300. Usually, the elderly often live with their families and contribute their pensions to the household budget. Meanwhile, from 1999 to 2002, China had returned a total of 7.6 million hectares of land to forests or grassland. Since 2002, the Xinghe government has paid RMB 1000 per hectare to farmers who returned their own farmland to forests or grassland each year. These state grants represent a significant and stable income in rural area, particularly when it is compared with the financial circumstances of many non-pensioners.
4.2. Determinants of Income Diversity
The interview data indicates that rural households derive income from farming mainly, which on average accounts for a 60% share of total household income, 66% in north hilly areas, 74% in mid plains areas, 52% in south mountain areas.
Table 2 shows that crop production is the most important source of income; more than 67% of farm households from the surveyed sample derive money from crops sale; subsidies for returning land to forests and grassland make up for nearly 64% of state grants a household receives; more than 53% of families have adult members working as wage laborers in urban areas; about 52% of households gain money from livestock fairs; about 38% of households derive income from potato production; and more than 12% of households earn money from vegetable sales.
Based on the income classification above, we describe statistical characteristics of a number of income sources.
Table 3 shows that the education level of the household head is significantly and positively related to the number of income sources, as studies [
19,
20] have explained that high education levels open the window to different economic activities. Good education means engaging in more diverse activities and yields greater chances of earning more money [
8]. The results also reveal that rural households with a small proportion of children and elderly people tend to diversify income sources, since a greater proportion of children of household members means that fewer family members are able to engage in agricultural production or migrant workers.
Based on marginal economic theory, in a farm of a particular size, the marginal product of additional family labor in crop production declines as the household size increases, which makes it likely that alternative sources of income would be attractive. This is why many surplus laborers engage in non-farm activities. The more working-age adults there are, the more likely it is that that household will have a range of skills that allows income diversity at the household level. In addition, the size of the farmland is also significantly and positively related to the number of income sources. Partly, more land means that those households have greater opportunities to diversify crop mixes and increase income from different crops. However, if a household has a large share of irrigated land that needs a large number of working-age adults to cultivate, it will be prone to fewer income sources from non-farm activities during the growing season.
4.3. Spatial Difference of Income Diversification at The Town Scale
In this sub-section, we investigate the regional spatial differences of the rural households engaging in different income generating activities. The field survey show that 27.2% of households have two sources of income, 33.3% of households have three sources of income, 19.6% of households have four sources of income, and 8.6% of households have five or more sources of income. The most common income combinations are CP, PLS, CPN, and CPNS in northern hilly areas; CV, CN, and CNS in mid plain areas; and CN and CLSN in southern mountain areas.
With Equations (1) and (2), we calculated the
Idiv and
Idep at the township scale of Xinghe (
Table 4). Like two sides of a coin, the trend of the value of
Idep shows a direction opposite to that of
Idiv. A higher value of
Idiv corresponds with a lower value of the
Idep. In other words, a high value of
Idiv and low value of
Idep means that a household can derive income from one source to compensate the loss of another caused by weather shock, crops or livestock market failure, or poor opportunities working as wage laborers.
Table 4 shows the value of
Idiv and
Idep at the township scale of Xinghe. The results show that DTY in a southern mountain area has the highest income diversity, with an
Idiv value of 1.189 and an
Idep value of 0.361; CSY in a northern hilly area has the lowest income diversity, with an
Idiv value of 0.507, and an
Idep value of 0.685. Sorted according to income diversity at the township scale, the descending order is DTY, TJM, ZG, TJ, ERD, HQ, DZ, WY, DKL, WGQ, SWS, ETZ, and CSY.
Table 4 also indicates that location plays an important role in the spatial differences of income diversity among those townships. Results show that mid plains areas have the highest income diversity, with an
Idiv value of 0.935 and
Idep value of 0.467; southern mountain areas are lower than mid plain area, with an
Idiv value of 0.812 and
Idep value of 0.535; northern hilly areas have the lowest income diversity, with an
Idiv value of 0.607 and
Idep value of 0.624. This indicates that the households in the central and southern areas have a more stable livelihood compared with those in the northern areas, and have more opportunities to compensate for the agricultural loss caused by drought with non-agricultural income. In other words, residents in the central and southern areas are more resilient than those in the northern areas of the county. According to the survey, the average number of income sources for households is 2.3 in northern hilly areas, 3.2 in mid plains areas, and 3.0 in southern mountain areas, respectively.
Furthermore, environmental attributes, such as precipitation, groundwater resources, and surface slopes, vary at different locations and determine the agricultural production activity types and crop types. In arid and semiarid regions, there is a significant difference in crop yield between irrigated land and rain-fed land. For instance, in the northern hilly area of Xinghe where there are large areas of rain-fed farmland, when the dry weather occurs in the growing season, it often leads to serious loss of agricultural production. Thus, it is difficult for towns like SWS, WY, WGQ, DKL, and CSY in the northern areas to obtain the same agricultural income as towns such as TMG and TJ, which have large areas of irrigated land in the mid-plain areas. According to field interviews, peasants in TJ can plant wheat, corn, potato, and some kinds of vegetables, while the options of peasants in SWS are limited to spring wheat and potato. Thus, the annual average farm income is above RMB 5000 in the central areas, contrasted to RMB 1500 in the northern hilly areas.
4.4. Why Rural Households Diversify Their Income Sources?
Economic reforms launched in the early 1980s have led to important changes in agricultural production activities in rural China. On the one hand, the Household Contract Responsibility System makes it possible for rural households to decide on their farm or non-farm activities by themselves. On the other hand, the emergence of market mechanisms encourages profit-oriented activities. Economic reforms have brought incentives and opportunities for rural households of large sizes to diversify both farm and non-farm activities. Thus, the interviews with household heads show that large households with few children and elderly members are also more likely to have a large share of income from farm or non-farm activities, especially in TMG or TJ in mid plains areas. In other words, it is true that the more working-age adults there are in a household, the higher share the non-farm income takes up.
Weather plays a key role in the household farm and non-farm activity diversification process. In Xinghe, drought is a major “push” factor that encourages rural households to turn to a more diversified portfolio of activities and thus strongly influences rural crop production, income, and welfare. Both farm and non-farm diversification can thus be seen as efficient mechanisms for households to reduce income risks caused by continuous drought. However, within the context of climate change, the uncertainty of weather events may make local rural households more likely to engage in new activities, especially non-farm activities.
Crop insurance is one of the most useful risk management tools for adapting to climate change [
21,
22,
23]. Since 2007, some large insurance companies (such as PICC) have provided crop insurance. Due to the crop insurance subsidy from the government, China’s crop insurance program has made unprecedented progress. However, current crop insurance focuses on flood disaster, not drought. In our field survey, local farmers said that they did not receive any drought crop insurance subsidies. Of course, some farmers in arid regions can receive crop drought insurance today. However, the implementation of drought crop insurance will face many obstacles in the future. The most important point is that drought occurs almost every year in arid and semiarid regions such as Xinghe. Meanwhile, one study [
24] has shown that the diversity of income sources significantly and negatively affects farmers’ willingness to participate in crop insurance. The field survey shows that, in Xinghe, most rural households believe that increasing the source of income is the best insurance to resisting drought.
Crop diversification is a drought risk management strategy. In Xinghe, the lack of household savings and agricultural insurance means that farmers are unable to smooth out consumption through
ex post mechanisms. Therefore, they can allocate household resources to secure a more stable income through diversifying farm or non-farm activities. This could be interpreted as drought risk management—a deliberate
ex ante strategy towards anticipating possible negative events and counteracting expected failures in various income streams in the future. For instance, based on the water demand for the varieties of different crops in the growing season, local farmers often adopt highly diversified crop planting strategies to reduce weather-related risks [
8,
25,
26], as Makate et al. pointed out that crop diversification is a viable climate-smart agriculture practice that significantly enhances crop productivity and consequently resilience in rural smallholder farming systems [
26].
Figure 2 shows that, in order to reduce drought risk, many local rural households in Xinghe divide their land holdings into several plots for different crops, such as spring wheat, naked oats, and potato, since the sensitivity of different crops on the environment is different. It can be concluded that mixed planting leads to diversified income sources for rural households. The seasonal use of labor in rural areas is very important for income diversification. Seasonal laborers working as wage laborers in urban areas are an important source of income for rural households. Non-farm activities could absorb surplus labor, help farm-based households spread risk, offer more remunerative activities to supplement or replace farm income, offer income potential during the agricultural off-season, and provide a means to cope with or survive crop failures caused by droughts. As
Table 2 suggests, 155 out of the 291 surveyed rural households have family members working as migrant workers in urban areas.