Discrete Dynamics in Nature and Society

Volume 2016, Article ID 4786090, 14 pages

http://dx.doi.org/10.1155/2016/4786090

## Ecoefficiency of Intensive Agricultural Production and Its Influencing Factors in China: An Application of DEA-Tobit Analysis

^{1}College of Public Management, Zhejiang University of Finance and Economics, Hangzhou 310000, China^{2}Department of Public Policy, City University of Hong Kong, Kowloon 999077, Hong Kong

Received 27 November 2015; Accepted 15 February 2016

Academic Editor: Kannan Govindan

Copyright © 2016 Heyuan You and Xiaoling Zhang. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

#### Abstract

The excessive use of inputs per unit of agricultural land poses a great threat to ecological sustainability. Using an input-oriented data envelopment analysis (DEA) model, this study analyzes ecoefficiency of intensive agricultural production in 31 provinces in China. The results show that the total efficiency of only six provinces can be considered fully efficient and that scale efficiencies are generally lower than technical efficiencies. Then, the spatial distribution of ecoefficiency is analyzed. The findings demonstrate that the provinces whose ecoefficiencies are maximal are primarily located in western China. The technical efficiencies in the western region are better than those in the eastern and middle regions. Imperfect scale efficiencies are distributed across all three regions. Furthermore, using the Tobit model, an analysis of the factors that influence ecoefficiency shows that the variables of farmland area per capita (FA), income per capita (IC), population per household (PH), and population burden coefficient (PB) have statistically significant impacts on total efficiency. The distinct effects of the variables on total efficiency are caused by their differential effects on technical efficiency and scale efficiency. Finally, suitable policies designed to improve ecoefficiency are proposed according to the local circumstances of each of the three regions.

#### 1. Introduction

Although it accounts for only 7 percent of the worldwide total, agricultural land in China feeds 22 percent of the world’s population. Grain production increased on average by 3% annually in the past 40 years in China [1]. The increased yields resulted primarily from greater inputs of labor, pesticides, chemical fertilizers, and energy [2, 3]. The increase in agricultural output has provided enough food to reduce hunger and improve human nutrition in China [4]. However, the excessive use of pesticides and chemical fertilizers creates various environmental problems, including soil erosion and water pollution [5, 6]. Meanwhile, these environmental problems pose great threat to human health and the sustainable household livelihoods [7–9].

Despite agricultural land being a highly valuable resource for achieving food security in China, high quality agricultural land is continuously converted into construction land due to rapid urbanization, industrialization, and poor government regulation [10–15]. Therefore, intensive agriculture has been widely advocated as a struggle to feed the increasing population considering the limited agricultural land [16]. The key point of intensive agriculture is the choice of a suitable input intensity that achieves an adequate desired production without increasing, and, more preferably, decreasing, environmental damage [17]. Ecoefficiency is a widely used indicator that describes the capability to achieve desired production and at the same time cause minimal environmental damage [18]. However, few studies have analyzed the ecoefficiency of agricultural production, especially with regard to intensive agriculture. Therefore, an ecoefficiency assessment is selected in this study to integrate ecological impacts into an analysis of intensive agricultural production in China and thereby inform certain changes in agricultural production that could help achieve sustainable farming.

This study uses a DEA approach to measure the ecoefficiency of intensive agricultural production in China. However, DEA has limited use for identifying the factors that influence efficiency. In order to conduct a further exploration of the factors that influence ecoefficiency, this study conducts the analysis over two stages. The first stage estimates the ecoefficiency of intensive agricultural production by DEA. The second stage analyzes potential underlying factors that influence the ecoefficiency by Tobit model. We specifically attempt to define the ecoefficiency which is suitable for this study; develop the Charnes-Cooper-Rhodes (CCR) and Banker-Charnes-Cooper (BCC) models for analyzing ecoefficiency; discuss input-output variables applied in the ecoefficiency assessment of intensive agricultural production and show the results of a case study pertaining to total efficiencies, technical efficiencies, and scale efficiencies in 31 provinces in China; analyze the influencing factors based on the Tobit model; and provide some policies designed to improve ecoefficiency in China.

#### 2. Literature Review

Ecoefficiency has been proposed as one feasible method for exploring sustainable development strategy [19]. Increasing literature has reported the applications of ecoefficiency into environmental studies. Ecoefficiency of 24 power plants in Europe and intensification scenarios for milk production in New Zealand were evaluated by treating pollutants as the inputs [20, 21]. Ecoefficiency analysis for regional industrial systems in China was performed using data from 30 provinces [22]. Ecoefficiency indicators were developed to design a framework for implementing cleaner production initiatives for the Canadian food and beverage industry [23]. However no agreement has been reached on the definition of ecoefficiency, since the ecoefficiency is a context-specific concept [20]. One important approach in the ecoefficiency literature is to define ecoefficiency as a ratio: ecoefficiency = economic value added/environmental damage [18]. But this definition is not suitable for DEA analysis since it neglects the resources which are used in the production process. In this paper, the ecoefficiency is defined as the capability that creates the highest economic value with lowest ecological damage. Therefore, it is the ratio of a weighted sum of the outputs (desired products, desired services, and ecological damage) to a weighted sum of the inputted resources [24]. One important implication of outputs is that the growth of desired products and services should be delinked as much as possible from harmful outputs including pollution and waste.

The methods used to assess the ecoefficiency include ratio calculation between economic value added and environmental damage, cost-benefit analysis, life cycle assessment, and data envelopment analysis [18, 30, 31]. Because the ecoefficiency is defined as the ratio including multi-input and multioutput in this paper, the methods are not suitable for assessing the ecoefficiency in this study except DEA. DEA which is a nonparametric methodology is an important method for evaluating the relative efficiency of decision-making units (DMUs) and is widely used to solve the efficiency related problems based on multi-input and multioutput production [32]. One of the obvious advantages of DEA is that it does not make any prior assumptions on the relationships between input and output variables. The complicated functional relationship between inputs and outputs in agricultural production process is not fully understood. Therefore, the DEA should be a suitable tool to assess the ecoefficiency of intensive agricultural production. Previous analyses performed using DEA models include a cross-country comparison of the ecoefficiencies of cement industries, an assessment of the ecoefficiency of pesticides, and a selection of transport modes based on ecoefficiency [33–35]. The ecoefficiency of the agricultural sector also has attracted the attention in past years. Ecoefficiency of olive farming at farm level in the rural areas of Andalusia in Spain was quantified by using DEA and pressure distance functions [36]. In addition, an input-orientated data envelopment analysis framework was developed to assess the ecoefficiency of 30 OECD countries [37]. Because rapid economic growth tends to correlate with environmental problems, increasing interest has been demonstrated in examining whether ecoefficiencies are sufficient for achieving sustainable development [18, 38].

#### 3. Methods

##### 3.1. Ecoefficiency Assessment Using DEA

The ecoefficiency assessment of intensive agricultural production involves two DEA models, including the CCR model and the BCC model, which defines efficiency as a ratio of a weighted sum of outputs to a weighted sum of inputs. The CCR model, which assumes constant returns-to-scale (CRS), measures the total efficiency of a DMU, and the BCC model, which assumes variable returns-to-scale (VRS), measures the technical efficiency of a DMU. The input-oriented or output-oriented CCR and BCC models, selected according to the purpose of production, are usually employed. The ideal intensive agricultural production level can be characterized as one that uses the lowest quantity of inputs to produce a given level of output. Therefore, an input-oriented CCR model and an input-oriented BCC model are employed to assess the ecoefficiency of intensive agricultural production in this study.

We assume there are homogeneous provinces, and each province consumes inputs and produces outputs. The input-oriented CCR model in ratio form can be summarized as follows [39]:where is total efficiency of the th province, and are the th input and the th output corresponding to the th province, respectively, and are the th input weight and the th output weight, respectively, and is a non-Archimedean infinitesimal variable.

Using a linear transformation, the input-oriented CCR model in dual form is expressed as follows:where is the multiplier of the th province, is the weight of reference set of the th province, is the input excess of the th input, and is the output shortfall of the th output.

The input-oriented BCC model which assesses the technical efficiency using linear program is obtained by adding the intercept in (1) and the convexity constraint [40]. Using a linear transformation, the input-oriented BCC model with the constraint in dual form is expressed as follows:

Finally, the measure of scale efficiency for ecoefficiency of intensive agricultural production in the th province is computed as the quotient between total efficiency and technical efficiency:where is technical efficiency of the th province.

##### 3.2. Tobit Model Analysis

Ecoefficiency has been criticized for excluding economic rationale and lacking connections to environmental policy instruments. This is because most of the previous studies treat DEA as a black-box to measure ecoefficiency. Explanation of differences in ecoefficiency analysis is important because inefficiency in environmental performance implies the existence of negative factors, which is difficult to justify in the traditional framework of environmental economics. It also connects ecoefficiency to environmental regulation and innovations in technology which is very relevant for policy purposes. The primary statistical method to analyze the linear relationships between variables is ordinary least squares (OLS). One important assumption for OLS is that the expected value of the residuals is zero. Violating this assumption, when applying OLS on censored or truncated data, it will not guarantee that the expected value of the residuals is necessarily zero. Therefore the parameter estimate is biased [41]. The values of ecoefficiency estimated by DEA are limited and lie in the unit interval . The Tobit model proposed by Tobin [42] is suitable for solving problems with limited dependent variables. Previous studies have applied the Tobit model to analyze explanatory variables to explore the determinants of truncated variables, such as environmental technological efficiency, health production efficiency, efficiency of worldwide railway companies, and efficiency of government spending on health [43–46]. Consequently, the Tobit model can be employed to help in the discovery of factors restricting ecoefficiency.

A maximum likelihood estimation is used to evaluate parameters in the Tobit model in this study. The standard Tobit model is expressed as follows [42]:where is the th province (DMU), is the latent dependent variable, is the limited sample value, is the matrix of the explanatory variables, , and is the error term, which submits to .

#### 4. Ecoefficiency Analysis of Intensive Agricultural Production

##### 4.1. Input-Output Variables

Input-output variables of agricultural production should be identified by using DEA to evaluate ecoefficiency. However, agricultural production is a complex system of plant and animal production with multiple inputs and outputs. Table 1 summarizes the inputs and outputs reported in previous studies [25–29]. The input variables and output variables selected herein reflect the purpose of this study. The ecoefficiency analysis of agricultural production aims at creating improved outputs and using fewer resources while discharging less pollution and waste. Intensive agricultural production is characterized by the high use of inputs relative to a given area of agricultural land. Therefore, the input variables that represent the characteristics of intensive agricultural production are defined as the inputs used per unit of agricultural land area. Although various inputs were selected in previous studies, this study only focuses on the five critical inputs in agricultural production: labor, agricultural machinery, pesticides, chemical fertilizers, and diesel oil. Consequently, the input variables chosen are labor intensity, agricultural machinery use intensity, pesticide use intensity, chemical fertilizer use intensity, and diesel oil use intensity.