Perception of Ecological Welfare from the Recycling of Agricultural Resources and Its Influencing Factors

The SLR is a special tool to solve regression problems, where the explained variables are discrete and the explanatory variables are continuous or discrete [11-13]. In this paper, the change in each of the three aspects of ecological welfare (air quality, water quality, and farmland quality) is scored by farmers as follows: “worse=1”, “the same=2”, and “better=3”. The scores thus obtained from a multi-class sequential data, which cannot be effectively analyzed by stepwise regression or multiple regression. Therefore, the SLR was introduced to systematically analyze the farmers’ perception of ecological welfare and its influencing factors:

${{y}_{i}}=\alpha +\sum\limits_{i=1}^{n}{{{\beta }_{i}}{{Z}_{ki}}}+{{\varepsilon }_{i}}$        (1)

where, y_i is the response of farmer i to the ecological welfare from the recycling of agricultural resources (y_i=1 means the response worsens, y_i=2 means the response remains the same, and y_i=3 means the response improves); Z_ki is the k-th factor affecting the response of farmer i; α is a constant; β_i is a coefficient; ε_i is a random disturbance reflecting the influence of hidden factors and statistical error [14, 15].

Let P(y=j|x) be the probability that the response falls on level j(j=1,2,3). Then, the probability that the response falls on or below j(j=1,2,3) can be expressed as P(y≤j|x)=P(y=1|x)+…+P(y=j|x). The cumulative probability that the response falls on or below j was subjected to logit transform:

$\log it{{P}_{j}}=\log it[P(y>j\left| x \right.)]=\ln \frac{P(y>j\left| x \right.)}{1-P(y>j\left| x \right.)}$       (2)

j=1,2

The logistics regression of sequential classification results can be defined as:

$\log it{{P}_{j}}=\log it[P(y>j\left| x \right.)]=-{{\alpha }_{j}}+\sum\limits_{i=1}^{p}{{{\beta }_{i}}{{x}_{i}}}$       (3)

j=1,2

Formula (3) can be rewritten as:

$P(y\le j\left| x \right.)=\frac{1}{1+\exp (-{{\alpha }_{j}}+\sum\limits_{i=1}^{p}{{{\beta }_{i}}{{x}_{i}}})}$        (4)

Then, level k was divided into two classes: {1,…,j} and {j+1,…,k}. Let p be the logarithm of the advantage of the cumulative probability that the response falls on the latter class over the cumulative probability that the response falls on the former class. Hence, the above model is called the cumulative advantage model.

The cumulative advantage model of the sequential classification results contains (k-1)+p parameters, in which α_i and β_i are estimated parameters. For any j, logitP_j is a linear function of the explanatory variable x: P(y≤1|x)< P(y≤2|x)<…< P(y≤k|x). It can be derived that α₁<α₂<…<α_k, where α_i is the division coefficient splitting the distribution. Then, the parameters were estimated by maximum likelihood method:

$P(y=j\left| x \right.)=P(y\le j\left| x \right.)-P(y\le j-1\left| x \right.)$

$\text{=}P(-{{\alpha }_{j}}+\sum\limits_{i=1}^{p}{{{\beta }_{i}}{{x}_{i}}}<u<-{{\alpha }_{j-1}}+\sum\limits_{i=1}^{p}{{{\beta }_{i}}{{x}_{i}}})$

$\text{=}\frac{1}{1+\exp (-{{\alpha }_{j}}+\sum\limits_{i=1}^{p}{{{\beta }_{i}}{{x}_{i}}})}-\frac{1}{1+\exp (-{{\alpha }_{j-1}}+\sum\limits_{i=1}^{p}{{{\beta }_{i}}{{x}_{i}}})}$                (5)

j=1,2

where, p_i is the probability of farmer i’s response to ecological welfare; α is a constant; β_i is a regression coefficient [16].

4.1 Farmers’ perception of ecological welfare from the recycling of agricultural resources

According to the results of descriptive analysis, the farmers’ perception of ecological welfare during the recycling of agricultural resources was divided into three aspects:

(1) Perception of air quality change

Among the 616 respondents, 29.36% held that rural air quality worsened in recent years, 44.57% believed that rural air quality remained the same, and only 26.07% recognized that rural air quality improved due to the recent transform and reuse of agricultural resources. Overall, the farmers had a low perception of ecological welfare in terms of air quality change. This means the recycling of agricultural resources has substantially changed the air quality.

(2) Perception of water quality change

Among the 616 respondents, 45.81% held that rural water quality worsened in recent years, 27.59% believed that rural water quality remained the same, and only 26.01% recognized that rural water quality improved due to the recent recycling of agricultural resources. Overall, the farmers had a low perception of ecological welfare in terms of water quality change. The recycling of agricultural resources has not markedly improved the water environment in rural areas.

Compared with air quality change, water quality change is closely associated with agricultural production. Therefore, the farmers are more sensitive to water quality change than air quality change. That is why 44.57% believed that rural air quality remained the same, and 45.81% held that rural water quality worsened in recent years. However, the proportions of farmers who thought that the air quality and water quality have improved were both below 30%, indicating that the farmers had a low perception of ecological welfare in terms of air quality change and water quality change.

(3) Perception of farmland quality change

Among the 616 respondents, 27.13% held that farmland quality worsened in recent years, 50.41% believed that farmland quality remained the same, and only 22.46% recognized that farmland quality improved due to the recent recycling of agricultural resources. Overall, the farmers had a low perception of ecological welfare in terms of farmland quality change. The recycling of agricultural resources has not reached a scale effect on improving the non-point source agricultural pollution.

To sum up, the farmers had relatively low perception of ecological welfare, although local governments have formulated many policies on the recycling of agricultural resources and rural eco-environment governance. There is still ample room to improve the ecological welfare of farmers.

4.2 Influencing factors of farmers’ response to ecological welfare

The multicollinearity of the explanatory variables was tested, using the calculation module of variance inflation factor (VIF) in SPSS 22.0. The results show that the explanatory variables were not linearly correlated. Furthermore, the model was found to have a high goodness-of-fit, for all explanatory variables passed the significance test at the 5% level. In addition, these variables also passed the test of parallel lines in SPSS 22.0.

On this basis, the SLR models were constructed for farmers’ perceptions of air quality change, water quality change, and farmland quality change, and analyzed by the Ordinal Regression module in SPSS 22.0. Based on the analysis results (Table 2), the factors that affect the farmers’ perception were discussed as follows:

4.2.1 Factors affecting farmers’ perception of air quality change

The regression results of model 1 show that, during the recycling of agricultural resources, the perception of air quality change has a significant negative correlation with education (significant at the 5% level), and a significant positive correlation with years of farming and satisfaction of nearby highways (significant at the 10% level).

According to the variable values in Table 1, when agricultural resources are being recycled, every level of increase in education reduces the score of air quality change by 0.675%; long years of farming and improved traffic conditions can effectively enhance the farmers’ favorable attitude to air quality change. Therefore, the farmers’ perception of ecological welfare can be greatly improved by training professional farmers and promoting infrastructure.

Meanwhile, the other eight variables showed insignificant impacts on the perception of air quality change. Their influence on farmer’s perception of air quality change can be neglected.

4.2.2 Factors affecting farmers’ perception of water quality change

The regression results of model 2 show that, during the recycling of agricultural resources, the perception of water quality change has a significant negative correlation with gender (significant at the 5% level) and education (significant at the 10% level), and a significant positive correlation with the distance to the nearest fair/market and satisfaction of nearby highways (significant at the 10% level).

According to the variable values in Table 1, when agricultural resources are being recycled, male farmers with relatively good education had low perception and recognition of water quality change; if there is a fair/market nearby and the traffic conditions are good, the farmers tend to perceive and recognize water quality change well. Therefore, the farmers’ perception of ecological welfare can be greatly improved by stepping up the construction of market infrastructure.

Meanwhile, the other seven variables showed insignificant impacts on the perception of water quality change. Their influence on farmer’s perception of water quality change can be neglected.

4.2.3 Factors affecting farmers’ perception of farmland quality change

The regression results of model 3 show that, during the recycling of agricultural resources, the perception of farmland quality change has a significant negative correlation with years of farming (significant at the 5% level), recycling infrastructure of agricultural resources (significant at the 10% level), and participation in specialized farmer cooperative (significant at the 5% level).

According to the variable values in Table 1, when agricultural resources are being recycled, the farmers who had engaging in farming for many years were sensitive to the changes in the “stereo” agricultural environment, and tend to perceive the worsening of farmland quality. In our survey, 59.35% of respondents replied that his/her village has no recycling infrastructure of agricultural resources; 86.00% of them answered that he/she had not participated in specialized farmer cooperative. This greatly limits the farmers’ perception of farmland quality change.

Meanwhile, the other seven variables showed insignificant impacts on the perception of water quality change. Their influence on farmer’s perception of farmland quality change can be neglected.

Table 2. The SLR results on factors affecting the perception of ecological welfare

Note: *, **, and *** mean the variable is significant at the levels of 10%, 5%, and 1%, respectively.

To sum up, the following results can be drawn from the regression results: the farmers’ perception ability of and sensitivity to the ecological welfare from the improved “stereo” agricultural environment could be effectively enhanced by cultivating professional farmers with relatively good education. Besides, the farmers could perceive ecological welfare much more effectively, if the processing infrastructure of agricultural resources (e.g. nearby roads) is improved.

The farmers’ perception of ecological welfare could also be greatly affected by personal variables like age, gender, part-time job (Y/N), and participation in training (Y/N), as well as external variables like nearby rivers or reservoirs (Y/N), and recycling infrastructure of agricultural resources (Y/N). However, these factors have not exerted an obvious effect on the perception of ecological welfare from the recycling of agricultural resources. This is because the Chinese farmers are generally poorly educated, most technical trainings are merely a formality, and rural infrastructure is still backward. Of course, the underlying reasons should be further researched by relevant subjects.