There is still controversy in the academic circle about the impact of the Grain for Green Program on the income of farmers. Some scholars believe that the Grain for Green Program may reduce farmers’ income. Based on the field investigation in Mei County, Shaanxi Province, Han and Shi (2010) pointed out that there were cases of local governments misappropriating and embezzling compensation funds during the construction of the Grain for Green Program, which reduced the family income. Zhang et al. (2019) used the CHIPS data in 2013 to explore the effect of the Grain for Green Program in the Yellow River and Yangtze River basins on farmers’ income reduction. The program mainly reduced farmers’ income by reducing labor supply and reducing human capital. Yang et al. (2019) used the data from the rural household follow-up survey sponsored by the Central Ministry of Finance and the Asian Development Bank. Based on the income level of farmers, it was found that the Grain for Green Program has significantly reduced the incidence of poverty, but it varied from year to year and from region to region, and the poverty targeting efficiency was low.

However, some scholars hold the view that the Grain for Green Program is helpful to help the poor and improve the income and welfare of farmers. Li and Tuo (2012) examined the positive impact of the Grain for Green Program on farmers’ income and welfare in Zhouzhi Mountain Area of Shaanxi Province. Liu et al. (2013) studied the dynamic impact trend of the Grain for Green Program in Wuqi County on the relative poverty breadth, intensity and depth of farmers, which first decreased and then increased. Wang and Ximing (2017) analyzed the positive impact of the Grain for Green Program on non-agricultural employment and farmers’ income during the “11th Five-Year Plan” period in 21 provinces in China.

Income is an important factor affecting residents’ well-being. Based on the survey data of Guangdong Province, Zhang and Cai (2011) found that the absolute income increase can significantly strengthen the residents’ well-being, and the absolute income level has an inverted U-shaped relationship with the residents’ well-being. Luo (2009) found that relative income can significantly enhance subjective well-being, and when considering the influence of absolute income and relative income on well-being, the absolute income variable coefficient is not significant. Therefore, the Grain for Green Program is likely to affect farmers’ subjective well-being by affecting their income.

A large number of documents show that the Grain for Green Program will improve the level of environmental quality and reduce environmental pollution. Based on the sediment data of 11 rivers in China, Deng et al. (2012) provided evidence that the Grain for Green Program could solidify soil and improve soil quality. Based on the carbon emission data, Chang et al. (2011) pointed out that the Grain for Green Program has significantly reduced the carbon emission in grassland and forest areas. Xiao (2014) has used the satellite database of the National Aeronautics and Space Administration (NASA) of the United States to show that the Grain for Green Program has significantly improved the air quality in the Loess Plateau region.

However, some scholars still hold negative opinions on the environmental improvement effect of the Grain for Green Program. It is believed that the Grain for Green Program will significantly worsen the water resources situation in arid and semi-arid areas, and large-scale construction of forest land in ecologically fragile areas will aggravate the regional water shortage (Shixiong et al., 2009). Chen et al. (2015) pointed out that the excessive scale of the Grain for Green Program will have negative impact on the community and the environment, including the shortage of food supply and the pollution and shortage of water resources.

And environmental pollution is the decisive factor affecting residents’ well-being. Yang and Zhang (2014) pointed out that air pollution can reduce residents’ health status, affect residents’ subjective mood, and thus inhibit residents’ well-being. Based on CLDS data, Linxiang and Zhang (2020) found that the higher the income, the stronger the inhibition effect of residents’ subjective air pollution on well-being, and the higher the income, the stronger the perception of individual subjective air pollution for the same environmental level, further strengthening the inhibition effect on well-being. Zheng et al. (2015) used CGSS data and found that subjective environmental pollution significantly inhibited residents’ well-being.

Therefore, the Grain for Green Program is likely to strengthen subjective well-being by reducing farmers’ subjective air pollution, subjective noise pollution and subjective soil pollution. However, the Grain for Green Program may also strengthen farmers’ subjective perception of water pollution, Shixiong et al. (2009) provide the statistical evidence in a northern part of China’s Shaanxi Province. Since there might be high demands for water as a consequence of the fast growth of pioneer species, such as hippophae rhamnoides, which may exacerbate soil water shortages, at least in the short term, especially in the arid and semiarid areas and further inhibit their subjective well-being.

Based on the above analysis, this paper puts forward two opposite assumptions:

Ha: Under certain other conditions, the Grain for Green Program inhibits farmers’ subjective well-being.

Hb: Under certain other conditions, the Grain for Green Program will strengthen farmers’ subjective well-being.

The data in this paper are derived from the CLDS2016 hosted by the Social Science Survey Center of Sun Yat-sen University. The sample covers the labor force aged 15–64 in 29 provinces and cities across the country (except Hong Kong, Macao, Taiwan, Hainan, and Tibet), including three levels of information: village residence, family, and individual labor force, which is representative of the country (Wang et al., 2017). The survey contains detailed information about the participation in the Grain for Green Program, life satisfaction and residents’ subjective well-being, which provides the basis for the study.

The core explained variable in this paper is farmers’ well-being. Interviewees were asked the question “On the whole, do you think your life is happy?” The answer measures of “very happy,” “happy,” “average,” “unhappy,” and “very unhappy” are assigned as 5, 4, 3, 2, and 1 in sequence by interviewees. The higher the value, the better the sense of well-being. The explanatory variable is the participation in the Grain for Green Program, and the CLDS questionnaire provides the situation at the village level. Among the control variables, personal characteristic variables include gender (1 infers male, 0 infers female), age, health (Interviewees were asked the question “How do you feel about your health now?” The answer measures of “very healthy,” “healthy,” “average,” “unhealthy,” and “very unhealthy” are assigned as 5, 4, 3, 2, and 1 in sequence. And it has been suggested that although there is some measurement error in self-rated health as a subjective indicator of health, it is still a good predictor of health status and is even more robust than objective measures of health (Poortinga, 2006), education level, political status (1 present party member, 0 present the other), religious belief (1 infers having religious belief, 0 present the other), family per capita income (get logged measured after calculated with household income and number of family members), marital status (1 infers getting married, 0 infers the other), debt (1 infers own debt, 0 infers the other), trust (very disagree, disagree, agree and very agree are assigned as 1, 2, 3, 4 in sequence). The regional characteristic variables include whether non-agricultural economy exists (1 infers work in non-agricultural occupation, 0 infers the other). It has been found that the socio-economic status, health statues and the high level of trust to society could increase the sense of social equity, which in turn impact the sense of well-being (Linxiang and Zhang, 2020; Ni, 2020). Table 1 shows the descriptive statistics.

Subjective well-being of the explained variable in this paper is a five-item ordered selection variable. Due to the incomparability of the distances between the adjacent options of the ordered variables, it is inappropriate to directly use the ordinary least squares method (OLS). This paper follows the processing method of Liu et al. (2018) and chooses the Ordered Probit model to estimate. This model has been used to study a series of hierarchical levels of residents’ subjective perception variables because it can accurately reflect the marginal changes of ordered variables. The subjective well-being variables in this paper are applicable to this model. The Ordered Probit Model is an extension of Probit Model, which was first developed by McKelvey and Zavoina (1975) and applied to deal with the measurement problem that the explained variables are in discrete order. The model assumes that there is a latent continuous potential variable happy* that can represent the explained variable (redistributing preferences and noting as happy_i) but cannot be directly measured to satisfy.

The explained variable happy_i in this paper is in discrete form and expressed by a nonlinear function F(⋅):

In the formula (3), k₁,k₂,k₃,k₄k₅ represent tangent points, and the latent variable h⁢a⁢p⁢p⁢yi* represents the interpreted variable happy_i in sorted discrete form according to different tangent point sizes.

Where happy_i, represents the subjective well-being of the i-th interviewee, green_i indicates whether the village where the i-th interviewee is located is participating in the Grain for Green Program. X_iindicates a series of village-level characteristics and personal characteristics control variables described above, and ε_i is a random interference term. We mainly focus on the coefficient α₁. If α₁ is positive, it indicates that the Grain for Green Program can strengthen the well-being of farmers. If α₁ is negative, it indicates that the Grain for Green Program can inhibit the well-being of farmers. If α₁ is not significant, the Grain for Green Program has nothing to do with the well-being of farmers.

As for the income mechanism, as the existing research indicates that there is an “inverted U” relationship between income and subjective well-being perception, the conclusion that the income mechanism is tenable cannot be reached only by studying the impact of the Grain for Green Program on income. It can be empirically tested through the intermediary effect model. The intermediate effect model needs to expand the regression of dependent variable to the explained variable (step 1), dependent variable to the intermediate variable (step 2) and intermediate variable to the explained variable (step 3) in turn. If the third measurement result shows that the coefficient of the dependent variable is significant, but there is a significant decrease in absolute value of the coefficient relative to the first measurement result, it can be concluded that the intermediate variable plays a part of the intermediate effect (Wen et al., 2004). And since the individual income is not provided in CLDS database, we use household income to calculate the income per capital data to present income level.

As shown in Table 3, under the condition that other control variables are consistent, column (1) is the regression result between the variable of the Grain for Green Program and the variable of well-being, column (2) is the regression result between the variable of the Grain for Green Program and the intermediate variable of family per capita income, and column (3) is the regression result between the variable of the Grain for Green Program and the variable of family per capita income on well-being. From column (2), it can be seen that the conversion of cropland to forest project has significantly reduced the per capita income level of households by 0.131 units, and it is believable at the level of 99%. Compared with the measurement results in column (1), the column (3)’s absolute value of the regression coefficient of the variable of the Grain for Green Program decreased from 0.051 to 0.039, and the significance of the regression coefficient decreased. Therefore, the Grain for Green Program significantly reduces the per capita income of households, and the variable of per capita income of households plays a part of intermediary effect in the process of reducing farmers’ well-being by the conversion of cropland to forest project.

As for the environmental mechanism, the literature review shows that subjective environmental pollution can significantly and directly reduce farmers’ well-being. If the Grain for Green Program can affect the subjective environmental perception, it can be concluded that the Grain for Green Program can improve farmers’ subjective well-being by reducing the subjective environmental pollution; Or the Grain for Green Program can reduce farmers’ subjective well-being by aggravating subjective environmental pollution.

In order to measure the level of subjective environmental pollution (or subjective environmental quality), this part is based on the questions in the questionnaire about air pollution, water pollution, soil pollution and noise pollution: “How serious do you think the following questions are in the place where you live?” “not serious at all,” “not too serious,” “more serious,” and “not serious at all,” which are assigned as 4, 3, 2, and 1 in sequence according to the answers provided by the farmers. The higher the value, the lower the subjective pollution degree of relevant environmental problems and the higher the subjective environmental quality.

As shown in Table 4, the Grain for Green Program has significantly improved the subjective environmental quality of air and soil and subjective noise-free perception, and significantly reduced the subjective environmental quality of water resources. This is consistent with the conclusion of the existing literature. That is, the Grain for Green Program intensifies the subjective perception of water pollution and reduces the subjective perception of air pollution, soil pollution and noise pollution. Therefore, the Grain for Green Program strengthens the subjective well-being of farmers by reducing the levels of subjective air, soil and noise pollution. At the same time, the Grain for Green Program can reduce the subjective well-being of farmers by strengthening the subjective perception of water pollution.

On the whole, the Grain for Green Program will ultimately inhibit the subjective well-being of farmers by reducing the subjective perception of air, soil, and noise pollution, intensifying the subjective perception of water pollution and reducing the per capita income of households.