Evaluation and Influencing Factors of Agricultural Eco-efficiency in Jilin Agricultural Production Zone Based on Super-SBM and Panel Regression Methods


 Agricultural eco-efficiency is a meaningful index that assess the agricultural sustainable development. Based on the super SBM-DEA approach incorporating agricultural carbon emissions and panel data regression, this study evaluates agricultural eco-efficiency and then investigates the influencing factors in agricultural production zone of Jilin Province. The empirical results show the following: (1) During the observation period, the average agricultural eco-efficiency exhibits a flat “M-shaped” fluctuating trend, a trend of fluctuant growth with phase characteristics, and the agricultural eco-efficiency of each county still has much room for improvement. (2) Significant spatial differences exist in agricultural eco-efficiency across counties. All of the studied counties, except for Nong’an, Huadian, Lishu, Yitong, Gongzhuling, and Qianguo, need to change the input and output structure to optimize agricultural eco-efficiency. (3) The panel data regression model verifies that the agricultural technology extension level, agricultural economic development level, agricultural industry structure, agricultural mechanization intensity and urbanization level have close correlations with agricultural eco-efficiency. (4) The research findings have important implications for policy makers formulating agricultural environmental policies in accordance with the local conditions of various counties.


27
The ecological and environmental issues caused by the continuous increase in the carbon 28 emissions accompanying industrialization and urbanization have become increasingly prominent, 29 attracting widespread attention from governments and scholars in various countries. Some countries, 30 such as the UK and Japan, have put into practice actions and plans to achieve energy conservation 31 and emission reduction. China, following the reform and opening up, has become the world's largest 32 carbon emitter and has long regarded energy conservation and emission reduction as a national 33 development strategy. In agriculture, China has made a significant achievement by feeding 20% of 34 the world's population with 7% of the world's arable land, reaching a high enough level to satisfy 35 the rising national demand for grain (Jin et al., 2019). Nevertheless, this achievement has actually 36 resulted in prominent ecological environment issues, such as the degradation of cultivated land and 37 organic material, the decrease in basic soil fertility, and the excessive consumption of agricultural 38 chemical material. Agricultural carbon emissions accompany the unreasonable agricultural 39 production activities, which makes agriculture the second main source of emissions in China. It 40 must be recognized that the traditional agricultural production model characterized by high inputs, 41 high consumption and low efficiency has become unsustainable. To respond to the severe carbon 42 emissions associated with agriculture, the Chinese government has applied the strategy of zero 43 growth in the usage of chemical fertilizers and pesticides, advocating the use of "grain storage in 44 the land" and "grain storage in technology" to tackle practical problems. Moreover, agricultural 45 subsidies have been provided to enhance agricultural production capacity and reduce agricultural 46 production risk and encourage farmers to adopt practices to protect agricultural resources. To 47 method and Tobit regression method are regarded as conventional instruments for analysis to 144 identify the influencing factors. These models provide significant reference by revealing the 145 temporal variation trends of agricultural eco-efficiency. 146 The previous research enriches the understanding of agricultural eco-efficiency, both 147 theoretically and practically. However, empirical studies on agricultural eco-efficiency that focus on 148 the main production area of agricultural products remain scarce, and spatially focused studies on 149 agricultural production zones from the perspective of major function oriented zones are lacking. 150 Accordingly, this study offers a potential contribution to the existing literature in two aspects. 151 Existing studies have concentrated on the spatial dimension of agricultural eco-efficiency at the 152 national level, provincial level, and city level, not the county level. Using the data available, this 153 study aims to examine the spatiotemporal characteristics of agricultural eco-efficiency at the county 154 level. In addition, in contrast with the traditional consideration of agricultural eco-efficiency that 155 ignores resource and environmental factors, the improved assessment of agricultural eco-efficiency 156 in this study accounts for the negative impact of resource constraints to accurately reflect the 157 performance of agricultural economic growth. As such, this study not only reveals the characteristics 158 of agricultural eco-efficiency over time and space but also estimates the potential influencing factors 159 to propose suggestions for policy makers and agricultural managers. 160 161

Analytical framework 162
Under the interactions among the economic, social and environmental systems, different 163 agricultural production conditions and human development factors are intertwined, complicating 164 the change process of the spatiotemporal pattern of agriculture eco-efficiency. One the one hand, 165 the ratio and scale of agricultural input and output directly affect agricultural eco-efficiency. 166 Agricultural practitioners, the actual implementers of modern agricultural production, determine the 167 amount and structure of input factors, such as land use structure, planting structure, farmland 168 management scale, farming methods, and level of production. Therefore, with the conversion 169 between input and output, changes in input-output structure can directly cause changes in 170 laborers to cities, agricultural policies and agricultural market conditions also significantly affect 174 agricultural eco-efficiency at the macro level. This study intends to explore the complex relationship 175 between agricultural production and agricultural eco-efficiency by considering both its direct and 176 indirect influences. To support this in-depth understanding, an analytical framework illustrating the 177 interactions between agricultural production and agricultural eco-efficiency is proposed in Fig. 1 The underlying principle of agricultural eco-efficiency is to create agricultural economic value 201 with less agricultural input while continuously reducing the effects on the ecology and natural 202 environment. Thus, the calculation of agricultural eco-efficiency integrates three dimensions: 203 production inputs, desirable outputs and undesirable outputs. Table 1 displays the agricultural eco-204 efficiency evaluation index system. Specifically, a conventionally used strategy is established to 205 measure the production factor inputs using the indicators of labor force, machinery, energy, 206 irrigation, chemical fertilizers, pesticide, and plastic membrane. The desirable outputs comprise two 207 types of agricultural output in the agricultural production zone, namely, agricultural output and grain 208 production capacity. For the undesirable outputs, this study employs the amount of agricultural 209 carbon emissions as the proxy measure. Referring to previous scholarly work (Tian et al., 2014;210 266.48(kg/hm 2 ) and 312.6 (kg/km 2 ), respectively. We multiply the emission coefficients by the 214 usage amount or acreage to calculate the total agricultural carbon emissions (Dubey and Lal, 2009;215 Huang et al., 2019). Table 2 presents a statistical description of the indexes used for assessing 216 agricultural eco-efficiency. 217 Table 1 Agricultural eco-efficiency evaluation index system.  (1) Agricultural machinery intensity (AMI). Because high-intensity agricultural machinery can 223 improve agricultural production but simultaneously have environmental impacts due to the 224 consumption of fossil fuels, the effect of agricultural machinery intensity on agricultural eco-225 efficiency is unknown. Thus, this study employs the ratio of the total mechanical power to the crop 226 sown area as the proxy variable of agricultural machinery intensity. 227 Where objective function ρ is the agricultural eco-efficiency, and its variation range can be improve the effectiveness of parameter estimation. Therefore, this method is widely used for 291 modeling economic problems. The model is defined as the following formula: 292 where Y it is the dependent variable, X it is the independent variable, β 0 denotes the constant, β 1 , β 2 , … 294 β n , β n+1 represent the regression parameters, ε is the random error, i represents the county and t 295 denotes the time. 296 To eliminate the heteroscedasticity of variables, we take the natural logarithm of the original 297 data for further conducting the panel data regression model. 298 lnAEE it =β 0 +β 1 lnAMI it +β 2 lnXMI it +β 3 lnFFMS it +…+β 7 lnUL it +β 8 lnRRIL it +ε it (4) 299 and in formula (3). Table 3 presents the descriptive statistics of the dependent and independent 301 variables used in this empirical study. 302

305
According to the abovementioned model specification, this study evaluates the agricultural 306 eco-efficiency using the super SBM-DEA approach that incorporates agricultural carbon emissions 307 and then investigates its influencing factors using the panel data regression method. 308 In 2013, agricultural eco-efficiency decreased slightly, and its spatial agglomeration feature 388 weakened, owing to regional differences in agricultural incentive policies. More than 15 counties 389 had a medium-low level of agricultural efficiency. A possible reason for the decrease in agricultural

Analysis of the panel data regression results 423
A Pearson correlation test between the eight independent variables is conducted before the 424 test results show that the scores of the correlation matrices are small, which sufficiently confirms 426 that all of the independent variables have weak correlations with each other. Therefore, the data for 427 these eight independent variables are considered reliable for examining the influencing factors in 428 the 26 counties studied in the panel data regression. 429 effects is more better than that the model with random effects. The FEM regression results of the 444 eight independent variables on agricultural eco-efficiency are shown in Table 5 and are further 445 The results of the panel data regression set out in Table 5

Estimation results analysis 456
The estimated coefficient of agricultural machinery intensity (AMI) is negative and significant 457 at the 1% level, implying that agricultural machinery intensity suppresses the improvement in 458 agricultural eco-efficiency as a whole. This finding is associated with the large amount of fossil fuel 459 such machinery consumes, which further increases the negative environmental impact in the 460 agricultural production process. Specifically, agricultural machinery intensity affects the 461 agricultural resource inputs and increases the agricultural production efficiency but also leads to a 462 large increase in the consumption of energy resources, such as diesel oil, with the consequence that 463 the agricultural machinery intensity does not promote the improvement in agricultural eco-464 efficiency. That is, higher agricultural machinery intensity means more diesel oil consumption and 465 more carbon emissions, which is not beneficial for agricultural eco-efficiency. Thus, the 466 development of agricultural mechanization must be controlled accordingly to the intensity and scale 467 of mechanization. can improve the agricultural eco-efficiency. Therefore, when the family farmland management scale 479 is large, the agricultural eco-efficiency is high. 480 Additionally, the correlation coefficient between agricultural technology extension level 481 (ATEL) and agricultural eco-efficiency is positive and significant at the 1% level, indicating that 482 improvement in the agricultural technology extension level tends to intensify the increase in 483 agricultural eco-efficiency. This is because agricultural professional and technical personnel can 484 guide farmers to implement environmentally friendly agricultural production methods, which is 485 conducive to driving productivity and optimizing the production process via the technology effect. 486 In addition, as displayed in Table 5, the estimated correlation coefficient is 0.093, passing the 487 significance test, which demonstrates that this indicator can sufficiently influence agricultural eco- improve farming methods and optimize agricultural materials for the high-quality development of 503 agricultural production. In addition, the counties with a high agricultural economic development 504 level are likely to achieve a balance between agricultural production and the ecological environment 505 with the growth of agricultural intensification and specialization, which is conducive to increasing 506 agricultural eco-efficiency. 507 The correlation coefficient of urbanization level (UL) indicates that this indicator has a 508 significant negative influence on agricultural eco-efficiency at the 1% level, which indicates that the 509 increase in urbanization level can hinder agricultural eco-efficiency. Moreover, the negative 510 correlation coefficient of urbanization level is larger than that of agricultural machinery intensity in 511 terms of the impact degree. This finding illustrates that among the selected variables, the 512 urbanization level is the greatest hindrance to agricultural eco-efficiency. The influence of the 513 urbanization level on agricultural eco-efficiency is similar to that of agricultural machinery intensity. 514 Since the eleventh five-year plan period, surplus labor has flowed between urban and rural areas, 515 accompanying rapid urban construction. As a consequence of the surplus labor transfer to urban 516 areas, urban development has also squeezed out the input of labor, capital and other factors required 517 for agricultural production, leading to changes in the employment structure. The laborers who 518 remain in the countryside have to use more agricultural machinery to compensate for the loss in 519 labor via the substitution effect, which can cause the deterioration of the agricultural ecological 520 environment. It is not difficult to understand that when the urbanization level increases, the 521 agricultural eco-efficiency may finally decrease. 522 Notes: *** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level.

Policy suggestions 545
According to the aforementioned contributing factors, differentiated suggestions are proposed 546 for policy makers. First, the results show that agricultural mechanization intensity constrains 547 agricultural eco-efficiency. Governments in the JAPZ should control the intensity and scale of 548 mechanization, eliminate agricultural machinery with high energy consumption and low production 549 efficiency and adopt new environmentally friendly technology to maintain the stability of the 550 agricultural ecological environment. Second, the correlation coefficient of technological progress is 551 not high, but it is large enough to have an impact on agricultural eco-efficiency, which indicates that 552 its key role in the improvement in agricultural eco-efficiency should not be neglected, although the 553 agricultural technology extension level exerts a slight effect on agricultural eco-efficiency. 554 Technical training and professional skills should be provided for farmers in the JAPZ to help them 555 better master relevant green, sustainable agricultural technologies. Third, the results show that the 556 family farmland management scale can accelerate the improvement in agricultural eco-efficiency, 557 which is conducive to the adoption of new agricultural technologies. A moderate scale of 558 agricultural eco-efficiency can relieve the pressure of high energy consumption on cultivated land. 559 Fourth, among the selected variables, the agricultural economic development level is the strongest 560 positive factor driving agricultural eco-efficiency, implying that the scale expansion and total 561 growth of the agricultural economy are still the key ways to promote agricultural eco-efficiency. 562 Hence, the continuous improvement of the agricultural economic development level is still one of 563 the vital ways to increase agricultural eco-efficiency. Taking ecological priorities and green 564 development as the guidance, the extensive agricultural production and management model should 565 be transformed, the development methods should be optimized, circular and ecological agriculture 566 should be developed, and the sustainable use of agricultural resources should be promoted to support 567 high-quality agricultural development in the JAPZ. Finally, the urbanization level exerts a negative 568 impact on agricultural eco-efficiency, indicating that ecology-oriented agricultural subsidies should 569 be improved and a high-efficiency compensation mechanism be established to stimulate the 570 enthusiasm for agricultural production in the JAPZ. 571 There is an urgent practical need for research on agricultural eco-efficiency under resource and 572 environmental constraints. This study intends to narrow the gap in the literature on agricultural eco-573 efficiency, and several limitations remain that deserve in-depth attention in future research. In fact, 574 agricultural eco-efficiency is also affected by natural factors, such as climate, soil properties and the 575 natural environment, which influence the input and output of agricultural production to some extent. 576 Due to the limited availability of data regarding natural factors, this study investigated only 577 socioeconomic factors. When more natural data are publicly available, the spatial dimension and 578 emission reduction targets ahead of schedule and achieve the green agricultural transition.