**4. Results and Discussion**

#### *4.1. Impact of Elite Households on the Area of Land Transferred Out*

The explained variable in models (1)–(3) are all land areas subcontracted to individuals for a fee, and the values are continuous variables. In order to avoid the influence of outliers, the key continuous variables in this paper are all made to shrink the tails (Winsor2), and the dependent variables take most of the values of 0. Therefore, Tobit model regression is mainly used. In order to test the existence of multicollinearity, OLS regression was also attempted and all models had variance inflation factors (VIF) less than 2, so the existence of multicollinearity was excluded.

Table 2 shows the results of the hypothesis for elite households and the area of land transferred out, with the key explanatory variable in model (1) being "whether or not the household is elite", and does not control whether it is a disabled family. The key explanatory variable in model (2) is "whether or not the household is incomplete" and does not control for whether or not the household is elite; Model (3) contains "whether or not the household is elite" and "whether or not the household is incomplete". Model (3) contains two dummy variables, "whether elite" and "whether incomplete".


**Table 2.** Benchmark regression results.

Note: Robust standard errors in parentheses. \*\*\* represents *p* < 0.01; \*\* represents *p* < 0.05; \* represents *p* < 0.1.

From model (1) in Table 2, we can see that the coefficient corresponding to whether it is an elite household is 0.242 and is significant at the 1% level, which statistically indicates that elite households significantly increase the area of land subcontracted to individuals for a fee. In contrast to model (2), replacing the variable of elite households with incomplete households changes the coefficient to 0.068 and the coefficient is no longer significant, indicating that incomplete households do not have a significant effect on increasing the area of land subcontracted to individuals for compensation. The sample in model (1) may

be both elite and incomplete households (for example, a household with a party member and an incomplete person), but not controlling for incomplete households leads to some bias in the results obtained. Hence, model (3) further controls for incomplete households based on model (1), and the results are basically consistent with model (1), with a coefficient of 0.240 for elite households and significant at the 1% level. From the results of models (1)–(3) in Table 2, it can be verified that elite households contribute significantly in the area of land subcontracted to individuals for a fee, and research hypothesis 1 is verified. This result indicates that the less privileged households are mostly at a disadvantage in terms of area of land subcontracted to individuals for compensation as these elite households may use your power to influence the land rental market. This gives a clear indication that China's agenda to eradicate poverty through rural revitalization should be strengthened in favor of less privileged households to make them self-sufficient to partake in the land rental market.

#### *4.2. Mechanism Inquiry*

Table 2 presents the effect of elite households on the area of land subcontracted for a fee but does not explore the inner influence mechanism. As mentioned earlier, this paper defines elite households as households with party members or village cadres in the household, which are dummy variables in the baseline regression. Exploring the influence mechanism of elite households on land transfer can be further divided into elite households, and groups with party members in the household, village cadres in the household, and households with a college education or above are included in different regression models to observe their influence on the area of land transferred out. Table 3 shows the estimation results after dividing the independent variables; the main explanatory variable of model (1) is whether the household has party members, the main explanatory variable of model (2) is whether the household has village cadres, the main explanatory variable of model (3) is whether the household has members with a college education or above, and the Tobit model is still used for estimation because the dependent variable 0 takes more values. The results are shown in Table 3.

Model (1) shows that households with party members will significantly increase the area of land subcontracted for compensation, while model (2) shows that households with village cadres do not affect the land transfer area significantly. Model (3) controls for the variable "whether the household has village cadres" based on model (1), the results still show that households with party membership significantly increase the area of land transfer, and research hypothesis 2 is verified. Party members and village cadres with political status indicate that their households are in the elite class of rural society, which may lead to the problem of monopoly in the price of "land transfer" compared with incomplete households or even ordinary households. In the context of rural revitalization in China, the willingness of grassroots cadres to transfer land has increased significantly, but will the area of land transferred also increase significantly in reality? The result of this paper is "no", so research hypothesis 3 is verified, which indicates that there are certain differences between party members and village cadres when they are faced with the decision of subcontracting land to individuals for a fee. The mechanism may be like this. Party members are mostly part-time farmers and have more social capital and relationships in certain regions. Their main business is mostly not related to agriculture, so they have a stronger willingness to transfer land because of the relatively strong social capital. It is helpful for them to gain an advantageous position in land transferring negotiation, thus forming a "seller's market" pattern of land transfer, and then further improve their willingness of expanding the land transfer area. For grass cadres, they are rooted in rural areas, forming a close interest linkage mechanism with rural production and management activities and grassroots governance. On the one hand, they are conscientiously engaged in grassroots governance; on the other hand, most of their work is also related to agriculture, and they are more closely connected with land resources and have more emotion with land and form a hard constraint, coupled with the heavy section of rural grassroots affairs embedded layer by layer, resulting in them, compared to party members (non-grassroots cadres), paying more attention to the social security function and the "livelihood support" function of land, so they are unwilling and dare not easily transfer their land on a large scale. As grassroots cadres, they naturally hope to realize scalable land transfer for better local development, but this may be only their good intention but not their ultimate practical action due to their identity, job responsibilities, and personal characteristics.


**Table 3.** Mechanism analysis.

Note: Robust standard errors in parentheses. \*\*\* represents *p* < 0.01; \*\* represents *p* < 0.05; \* represents *p* < 0.1. In this part of the regression, in order to obtain the net effect of party members and village cadres on the dependent variable, models (1) and (2) were repeated. In the repeated model (1), the samples whose family members had village cadres were excluded. Model (2) excludes the regression of samples whose family members have party members, and the results are consistent with the above table. Due to space reasons, it is not reported.

#### *4.3. Robustness Test*

Benchmark regression shows that elite households will significantly increase the area of land subcontracted to individuals for a fee. To demonstrate that the results of this paper are robust, a series of robustness tests are conducted below, including replacing the measures, transforming the estimation model, and changing the estimation sample.

#### 4.3.1. Replacement of Measurement Index

In the baseline regression section, the explanatory variables are dummy variables, and households with party members, cadres, or university and higher education in the household are defined as elite households (FM\_str1). In the robustness test section, the explanatory variables are replaced with continuous variables, and elite households (FM\_ str1) are replaced with elite degree (Elite). Similarly, the incomplete family (FM\_str0) is replaced by the incomplete degree (Incomplete). The obtained results are shown in Table 4 below. In Table 4, model (3) is the full variables result, model (1) is the result without the variable of degree of incomplete relative to model (3), and model (2) is the result without the variable of degree of elite relative to model (3). Other control variables were added to all models in Table 4, and the control variable results are generally consistent with the

baseline regressions; therefore, the results are not reported. In Table 4 below, the coefficient corresponding to the degree of elite remains significant at the 1% level, indicating that as a household's degree of elite increases, it significantly increases the area of land subcontracted to individuals by that household for compensation, which is consistent with the research hypothesis and the results of baseline regression in this paper, including that the findings are relatively robust.


**Table 4.** Substitute explanatory variable metrics results.

Note: Robust standard errors in parentheses. \*\*\* represents *p* < 0.01; \*\* represents *p* < 0.05; \* represents *p* < 0.1.

#### 4.3.2. Transformation Estimation Method

In the above, Tobit models were used for estimation in all cases due to the presence of broken tails. However, for a continuous variable such as the area of land transferred out, the least squares (OLS) method is also used as a succinct method. In order to investigate whether the estimation results of this paper may change depending on the estimation method, the stability of the results of this paper is analyzed using different estimation methods. Table 5 shows the results of least squares estimation, and the coefficients and significance of the results are completely consistent with Table 2, indicating that the results estimated in this paper do not change, due to the change in estimation methods, and the results are robust and reliable. In addition, to explore whether there is multicollinearity, the posterior multicollinearity inflation factors (VIF) of the estimated results of models (1)–(3) are all between 1.01 and 1.39, so there is no multicollinearity.

**Table 5.** Transform Estimation Method Results.


Note: In parentheses are the robust standard errors; \*\*\* represents *p* < 0.01; \*\* represents *p* < 0.05; \* represents *p* < 0.1.

#### 4.3.3. Changing the Estimation Sample

Although the baseline regression and model (3) in the robustness test section control for the variable "whether or not the household is incomplete", this is not as intuitive as directly using the sample with incomplete households excluded. For this reason, the samples used in the next regressions in this paper are directly excluded from the sample of incomplete households and analyzed using the Tobit model used in the baseline regression. The results in Table 6 show that the coefficient corresponding to elite households is 0.224, which is significant at the 1% level, and its results are basically consistent with the baseline

regression, once again proving that the regression results in this paper are robust and reliable, and will not change due to the sample transformation.


**Table 6.** Transform Estimation Sample Results.

Note: In parentheses are the robust standard errors; \*\*\* represents *p* < 0.01; \*\* represents *p* < 0.05.

Based on different ways of robustness testing, the result of this paper is stable. Elite households significantly increase paid subcontracted land area.

#### **5. Conclusions**

#### *5.1. Simple Conclusions*

Whiles many are abandoning farmland for non-farm activities, many nations such as China are putting down strategies to ensure sustainable agricultural development, hence projecting a land transfer agenda. To help policymakers in their decision, this paper empirically analyzes the effect of family structure heterogeneity on the area of land transferred out in the current land transfer process in China, using Tobit regression, OLS regression, and a series of robustness tests, based on the conceptual definition and categorization of family structure, with data from the China Household Income Survey (CHIP)2013. Further, this paper discusses the effect of intra-elite household differentiation on land transfer area by subdividing elite rural households into those with party member status (non-grassroots cadres) and those with grassroots cadre status and finds that households with party member status have a significantly more individual land area in paid subcontracting, while households with grassroots cadre status do not have a significant effect on land transfer area. The possible explanation is that many party members (non-grassroots cadres) are part-time farmers who enjoy more social capital and more social relationships in a certain area and are more engaged in non-farm work, so they are more willing to transfer their contracted land. In addition, because of their relatively strong social capital, they have an advantageous position in the negotiation of land transfer prices and form a "seller's market" pattern of land transfer, and then want to expand the area of land transfer strongly. Meanwhile, grassroots cadres are more tied to the land and their reliance on the land is more obvious due to the social security and livelihood, and they are reluctant to easily transfer the extra land compared to the families with party membership.
