Does Internet and Information Technology Help Farmers to Maximize Profit: A Cross-Sectional Study of Apple Farmers in Shandong, China

Abstract

The adoption of Internet and Information Technology (IIT) in organizations has been growing at a staggering pace. In agriculture, IIT has emerged from the prospects of modern agricultural structure, which profoundly bring revolution in the way of agribusiness. While the impacts of IIT for selecting productive sales and marketing channels is evidenced by the substantial literature in the field, there is a crucial research scope of inclusive analytical views, especially in an economics context. The prime objective of the article is to assess the impacts of IIT for choosing a productive sales and marketing channel. Moreover, we tend to find whether the usage of IIT can eventually foster the profitability of the farmers. The empirical set of data is collected from a cross-sectional survey conducted in Shandong province, China. We utilize the Ordinary Least-Squares (OLS) regression, propensity score matching (PSM), and Heckman’s two-stage regression approaches to craft the findings. The greater extent of the use of IIT, the more significant and positive the impact of agricultural income is. After using the Heckman regression and PSM model, IIT’s use significantly increases the efficiency for selecting the sales channel, and the impact on agricultural income is also prominent (around 40%). We also find that the supporting and nonagricultural income exceeded 30%. Finally, the outcomes of the study reveal significant positive impacts for selecting productive sales and marketing channels. On the basis of these findings, it is suggested that the government and relevant departments should strengthen the construction of agricultural information platforms and websites. Authorities should also extend the training facilities of fruit farmers regarding the use of IIT, which could be useful to boost the capability of fruit farmers to develop markets and promote the value chain.

1. Introduction

The proper utilization of Internet and Information Technology (IIT) becomes part and parcel within a wide span of sectors. Especially in trade and commerce, it becomes inevitable, as IIT helps one to collect, store, and disseminate information at a staggering pace [1,2]. Currently, with the swift advances of globalization and enormous competition, the economic impact of IIT on rural society has also been continuously extended. This is particularly manifested because IIT provides the rural society with several distinct forms of innovation, particularly production and distribution [3,4]. On the one hand, the innovation of the quality of a supervised model of the entire agricultural production and value chain is represented by the Internet of Things (IoT) and big data [5,6]. Simultaneously, it provides innovative market circulation channels for rural society, which effectively promote the realization of the industry’s actual situation, upgrades the farmer’s income, and fosters economic development within rural areas [7,8,9]. The widespread use of modern IoT, powered by the internet and mobile phones, gives more opportunities to participate in the market by reducing the cost of the information search for large and small farmers and increasing the market transition. Interestingly, it could be vital for the background of supply-side structural reform and rural revitalization strategy initiated by the government [10,11,12]. Therefore, the widespread introduction of IIT tools will possess a vigorous impact in breaking down the information dilemma for farmers, reshaping the rural society and economy, and improving market access capabilities [13,14]. To this end, using the IIT to foster renewable agricultural development and overall rural development, which allows the majority of farmers to join in the in-depth economic distribution of resources and share the benefits of market-oriented reforms of further discussion.

While the implementation and generalizability of IIT within agriculture is a significant indicator of agro-agricultural modernization [15,16], the swift growth of IIT in the modern Chinese farmland production system could be a potential for upgrading conventional farming, especially for developing nations [17]. The prospects of successful agricultural initiatives are crucial to fulfilling the purpose of ensuring the smooth transition of information and the competitive ability of agribusiness farms [18]. It is obviously disappointing if the flow of information lags and comes too late to cope with. In such situations, communication becomes cohesive for farmland management [19,20,21]. Especially, it could be profound for high-value fruits like the apple. In China, sales and marketing involve a high level of intermediaries, and it is critical for apple farmers to choose the appropriate sales channel [22]. As per the usage of IIT, Chinese apple farmers accomplish commercial advancement by exploring a solution to render the products, which become quite meaningful for the transition of the farmer’s livelihood.

In practice, the interactions of IIT within the context of selecting sales and marketing could be crucial, as the agricultural product price is dynamic, which depends not only on the quality of the product but also on several externalities like storage facilities, timing, knowledge, and climatic situation. Conceptually, it is more apparent as every farmer wants to maximize profits [23]. Although the transition cost of IIT is somehow easier to consider, it is also challenging to assess the additional opportunities for improved revenues. However, scholars generally believe that IIT is conducive to promoting agriculture’s industrial upgrading [24,25,26] and eventually facilitate growth in income. In particular, the application and promotion of IIT embodied by mobile phones and computers in developing countries have significantly reduced information search costs for farmers and overcome information barriers [27,28]. Shimamoto et al. [29] believed that to maximize profits, farmers must fully grasp the market dynamics and information and decide on the production and sales of agricultural products according to their conditions to obtain the maximum benefits. Deichmann et al. [30] combed the influence of IIT in emerging nations and argued that internet information technology overcomes information occlusion that prevents small farmers from entering the specialized markets. Nakasone et al. [31] provided a brand new service promotion method and improved agricultural supply chain management, which is represented within IIT in agriculture. Zanello [32] showed that getting market information via mobile phone and the internet possesses a crucial impact on sales models. Interestingly, Aker and Ksoll [33] proposed that the use of information technology by smallholders possesses progressive influence on the acquisition of agricultural information and emphasized that, to some extent, learning how to use it is more important than simply owning these modern information technologies. Li and Huang [34] believe that the basis of rural socio-economic development is the upgrading of rural industries. Internet information technology has led to huge changes in business models in various fields of the agricultural, industrial chain, such as agricultural production, distribution, marketing, and services. At present, Chinese society is going through a crucial phase of socio-economic development and rapid transition of technological improvement [35,36]. The transition comprises an outstanding importance of market dynamics and proper utilization of valuable resources supported by the government [37,38]. Technical innovation powered by IIT acts as a blessing as regards facilitating fundamental rural economic and social development [39,40].

However, from the perspective of promoting farmers’ income, some scholars believe that the acquisition of information can change the distribution and production structure of production factors, increase agricultural productivity, reduce unnecessary intermediate links, and promote a substantial increase in farmers’ economic returns [41,42]. The research of Muto and Yamano [43] showed that the utilization of IIT could assist farmers in accessing and improving the market, facilitate resource efficiency, and promote the convergence of agricultural product prices in off-site trading markets, particularly among emerging nations. Zhang et al. [44] analyzed the potential ways of agricultural modernization to promote farmers’ income growth, including the advancement of agricultural production structure, the transformation of employment structure, and the sharing and use of agricultural resources. However, Tandi Lwoga et al. [45] revealed that farmers are usually required to have basic internet usage knowledge and skills, and at the same time, they must be able to bear various initial costs for fostering better access of agricultural information. Seemingly, Zhao et al. [46] specifically examined the impact and mechanism of internet regarding capturing farmers’ revenue via effective pesticide usage and revealed that usage of internet can foster income in rural areas. The outcomes portrayed show that increasing the practice of IIT in rural areas has a more obvious effect on farmers’ income. Although the prospects offered by these phenomena appear widely visible, there are also ample controversies and discussion about how and to what extent the utilization of the IIT possesses the agricultural sectors in terms of better access of market information [47,48].

Enhanced transmission of knowledge is broadly consented to have a beneficial impact on the development of the agribusiness industry and interpersonal capabilities. However, it is also problematic as well as expensive to capture the real-time benefit, especially, the dynamic industries like agribusiness, where the prime actor usually resides in rural areas [49,50]. The continuous use and promotion of network technology in rural areas have opened up space for further theoretical research in this field. On the one hand, most of the existing research focuses on whether farmers have modern IIT and lack paying attention to the impact of different degrees of farmers’ use of IIT on sales channels and sales revenue. Moreover, existing studies also mainly focus on grain farmers as the research object, and there is a limited study that can be traced on high-value agro-products such as fruits. Therefore the impact of IIT on selecting effective sales and marketing of high-value agricultural products like apples would foster high research value.

2. Theoretical Analysis, Data Source and Model Setting

2.1. Theoretical Prospects

First of all, Chinese farmers are mainly small- and medium-sized farmers, as shown in Figure 1. According to statistics, as of the end of 2016, small- and medium-sized farmers with less than 50 acres still accounted for 97% of the overall figure of growers. The use of IIT can broaden the access of small and medium-sized farmer’s channels, which in turn, affects their choice of sales channels. Compared with self-selling, cooperatives and intermediaries have broader information channels and stronger organizational capabilities, and have stronger bargaining power, which can help increase the sales of agricultural products [51,52,53]. As selling prices fluctuate, reducing the cost of information search can help the small- and medium-size farmer to increase the sales income [54,55]. Secondly, the use of IIT can help the majority of small- and medium-sized farmers to grasp the market dynamics in time and broaden the market scope [56,57] and thereby offer improved sales markets at home and abroad and promote the substantial increase in sales prices and sales volume of advantageous agricultural products, thereby achieving a significant increase in sales revenue [58,59]. Third, according to transaction cost theory, the use of IIT by small- and medium-sized farmers can help to reduce the sales process of agricultural products, shorten the time for bargaining, reduce the price of bargaining, and foster decision-making, thereby reducing the cost of agricultural products and increasing sales revenue [60,61,62,63]. Fourth, the use of the internet can affect farmers’ access to information and improve farmers’ labor skills and non-agricultural employment ability, making up for the lack of education level. Fifth, the use of the internet helps to expand the social network and employment channels of farmers, so as to enhance the possibility of non-agricultural employment and income levels.

2.2. Data Source

The study was conducted by a research team and was helped by the National Apple Engineering Center of Shandong Agricultural University in the collection of the empirical data from the major apple producing areas from September to November 2019. The survey is divided into two steps: In the first step, the research team first selected Guanshui Town, Muping District, Yantai City, and Shiliang Town, in Longkou City as pre-survey townships, and then randomly selected six apple farmers from each township for the pre-survey. The survey content includes the basic situation of fruit farmers, whether or not fruit farmers use IIT, related income and expenditure, and so forth. After getting the feedback collected via pre-assessment, the questionnaire was modified and improved. In the second stage, a screening criterion was conducted for the areas where apple planting is concentrated and adjacent, covering nine apple counties (cities) and districts under the jurisdiction of Yantai, Zibo, and Linyi. The survey used a combination of probability and size proportional sampling (PPS sampling) and simple random sampling tactics. A total of seven towns were selected from each sampled county (city, district), and then four villages from each selected town were randomly selected. After that, we randomly selected two fruit farmers from each surveyed village. The respondents mainly selected the head of the household or the main family members engaged in apple cultivation. A sum of 500 questionnaires was allocated in the survey. Through interviews with the investigators, filling out the questionnaires, and reviewing them, the total number of valid questionnaires (with full information) was 471, and the effective rate was 94.2%.

Table 1. Descriptive statistics of variables.

Variable Name	Meaning and Variable Assignment	Mean	Standard Deviation
Dependent Variable
Y1 self-sale	Through self-selling channels (1 = yes, 0 = no)	0.620	0.486
Y2 middlemen sales	Through middlemen sales channels (1 = yes, 0 = no)	0.701	0.551
Y3 cooperative sales	Through cooperative sales channels (1 = yes, 0 = no)	0.493	0.504
lnY4 agricultural income	Calculated by the gross sales revenue of Apple in 2017 (yuan) (Take the logarithm)	10.122	0.838
lnY5 supporting income	Household supporting income in 2017 (yuan) (Take the logarithm)	10.134	0.828
lnY6 total household income	According to total household income in 2017 (yuan) (logarithmic)	10.44	0.839
$W_i$ Whether to actively use Internet information technology	Whether to actively utilized modern Internet information technology to acquire agricultural information: non-active use = 0; active use = 1	0.592	0.496
$W_{i1}$ (Fruit farmers less than 60 years old)	Whether to actively utilized modern Internet information technology to acquire agricultural information: non-active use = 0; active use = 1	0.536	0.531
$W_{i2}$ (Fruit farmers aged 60 and above)	Whether to actively utilized modern Internet information technology to acquire agricultural information: non-active use = 0; active use = 1	0.600	0.490
Intervention Variables
lnX0 Use of Information Technology	The logarithm of the total cost of annual mobile phone communication by farmers, reflecting the degree of Internet information technology use (yuan) (take the logarithm)	25.824	6.803
Control Variable
Basic characteristics of respondents
X1 gender	Female = 0; male = 1	0.369	0.482
X2 age	Respondents’ age (years of age) in 2017	51.098	8.755
X3 education	Respondents’ years of education (years)	7.642	5.223
Production and management characteristics
X4 planting scale	The average apple planting area from 2014 to 2017	4.218	0.483
X5 years of planting	How many years have the apples been grown	27.325	6.067
X6 Whether to join a cooperative	Whether to join the cooperative: not join = 0; join = 1	0.707	0.664
X7 Specialization	The proportion of apple production income to total household income in 2017 (%)	77.363	4.731
X8 Have you ever gone to work	“Have you ever worked outside or recently?” No = 0; Yes = 1	0.607	0.489
X9 Ease of use of communication technology	“Do you think the use of communication technology is convenient?” Very inconvenient = 1; inconvenient = 2; average = 3; convenient = 4; very convenient = 5	2.270	1.255
X10 Planting Technology Training	“How often have you recently or recently participated in training in planting technology?” No participation = 1; Rare participation = 2; Frequent participation = 3	2.532	0.704
X11 Fertility of orchard soil	“How fertile is the orchard plot?” Poor = 1; Fair = 2; Good = 3; Very good = 4	2.382	1.076
X12 Orchard fineness	“How scattered is the orchard plot?” Concentration = 1; More concentrated = 2; Dispersion = 3	1.467	0.677
X13 How far the orchard is from the market	“How far is the orchard from the market?” Near = 1; Closer = 2; Farther = 3	1.473	0.822

indicates the fundamental attributes of the surveyed participants. The basic analysis shows that more than 60% of the respondents are women, the average age is over 50%, and most of them have education levels below junior high school. The average grower scale of fruit farmers is below 5 acres, but they are specialized. To a higher degree, more than 60% of the fruit farmers’ apple production accounted for above 80% of the overall family income, and more than 70% of the sample fruit farmers joined professional agricultural cooperatives (Table 1). Overall, the sample fruit farmers are well represented.

2.3. Model Setting and Variable Selection

In terms of variable selection, this article uses three different sales channels: self-sale, middlemen, and cooperatives, as suggested by Sheng and Lu [64] and Jiang and Sun [65]. We also investigate the household agricultural income, supporting income, total income, and whether to vigorously utilize IIT to obtain overall market information as the dependent variable, as suggested by Xu et al. [66]. The degree of use is the core explanatory variable, and the basic characteristics, production, and management characteristics of the interviewed fruit farmers are selected as control variables, as suggested by Valencia et al. [67]. The specific variables and descriptive analysis are shown in Table 1. The least-square method (OLS) was first adopted for analyzing the influence of fruit farmers’ IIT usage on sales and marketing network choice and income. The equation is as follows:

$$Y_i*\mathrm{or}\ln Y_i\mathrm{or}{W}_i={\beta }_0+{\alpha }_1{X}_i+{\mu }_i$$

(1)

where $Y_i*$ indicates that the investigated fruit farmers have different choices of the three sales channels. However, $\ln Y_i$ is the logarithm of the agricultural income, supporting income, and total income of the ith number of a researched fruit farmer family. $W_i$ indicates whether the fruit farmers actively use modern IIT to obtain supporting information. $X_i$ is the observable characteristic variable affecting fruit farmer income for the ith number of fruit farmer families. It should be noted that because the sample fruit farmer is relatively older, has a lower cultural level, and has more traditional living and consumption methods, the use of IIT is mainly aimed at the production and marketing information search presented through mobile phone internet access and telephonic consultation. Therefore, this paper uses the total annual communication costs $X_0$ of farmers to replace the use of IIT as intervention variables, and the remaining $X_i$ are control variables, and ${\mu }_i$ represent random error terms.

This article also analyzes the influence of fruit farmers’ use of IIT on the choice of sales and marketing networks and household income. One of the important issues that cannot be ignored is that fruit farmers’ different behavior choices and differentiated part-time behaviors lead to the possibility that fruit farmers can obtain agricultural and supporting income, and the process of fruit farmers’ income behavior selection can be divided into two processes: whether to choose to contribute in the progression of apple farming labor and join in the course of obtaining agricultural labor income derived from agricultural labor, or choosing supporting labor based on working outside and obtaining supporting income. Therefore, only by observing whether the fruit farmer participates in the choice of agricultural labor can we further examine the corresponding income subsequently obtained. Therefore, it is inevitable that there could be sample bias issues. In order to solve this kind of bias due to selection errors as much as possible, Heckman’s two-stage regression model can be used for regression processing to verify the impact of fruit farmers’ use of IIT on their agricultural income and supporting income, as suggested by Puhani [68] and Winship and Mare [69]. Heckman’s two-stage regression model mainly involves two equations, namely the selection equation and the result equation [70]. First, use the selection equation to estimate how fruit farmers choose agricultural labor and supporting labor. Then, use the selection equation to calculate the inverse mills and different choices of labor participation and other control variables as independent variables. Next, we use the result equation to estimate the use of IIT. The effect on the agricultural and supporting income of fruit farmers is calculated as follows:

$$\begin{array}{l}\mathrm{selection}\mathrm{equation}:Y^{*}=Z_i\eta +{\mu }_i\mathrm{if}{Y}^{*}>0,w_i=1,\\ otherwise\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{w}_i=0,\mathrm{Prob}\left(w_i=1|Z_i\right)=\phi \left(Z_i\lambda \right)\end{array}$$

(2)

$$\mathrm{r}esultequation:\ln \left(Y_i^{*}|w_i=1\right)=\gamma X_i+{\omega }_i$$

(3)

Among them, (2) is the choice equation, (3) is the result equation, $w_i$ represents the decision-making choice, $w_i=1$ represents the choice of agricultural labor, $w_i=0$ represents the choice of supporting labor; $Y^{*}$ is a latent variable, $\ln \left(Y_i^{*}|w_i\right)$ is the logarithm of agricultural income and supporting income obtained by fruit farmers; $X_i$ is an independent variable observed by fruit farmers’ family i that affects agricultural income and non-total income; $Z_i$ is regarded as the vector of latent factors that concludes the result of the selection equation; $\eta$ and $\gamma$ is two groups of factors to be assessed, ${\mu }_i$ and ${\omega }_i$ is the residual span of the two equations, $\phi$ is the standard collective distribution function, which obeys bivariate standard distribution of N₂ (0, 0; σ, ε, 1; ρ).

Table 1 shows that the use of IIT by fruit farmers is a choice made by fruit farmers according to their resource endowments. It is not casual behavior; however, it could be an assessment of self-selection. The extent to which fruit farmer’s use IIT is usually determined by their own and family characteristics and production and management characteristics, and these characteristics will inevitably have an impact on the relevant income they receive, which will lead to endogenous problems in the estimation of model-related income. An attempt will be made to utilize the propensity score identical technique to resolve the issues of bias caused by self-selection. According to Rosenbaum [71] the tendency score matching method, the average effect ATT of the treatment group can be expressed as:

$$ATT=\frac{1}{N_1}{\displaystyle {\sum }_{i{D}_{i=1}}\left(y_{1i}-y_{0i}\right)}$$

(4)

The basic steps are as follows: First, use the relevant income $y_{1i}$ obtained by the active selection of IIT by the fruit farmers (processing group) and the corresponding income $y_{0i}$ of these fruit farmers if they do not use the related information of IIT (control group) and related variables X_i such as D_i to estimate the use of fruit farmers probability scores and then match the propensity scores based on the probability and control the standardized deviation of each component of the relevant variable X_i:

$$\frac{\left|{\overline{x}}_{\mathrm{treat}}-{\overline{x}}_{\mathrm{control}}\right|}{\sqrt{\left(s_{x,\mathrm{treat}}^2-s_{x,\mathrm{control}}^2\right)/2}}$$

(5)

where ${\overline{x}}_{\mathrm{treat}}$ and ${\overline{x}}_{\mathrm{control}}$ represent the sample mean of the processing cluster and the control cluster; $s_{x,\mathrm{treat}}^2$ and $s_{x,\mathrm{treat}}^2$ represent the sample variance of the processing and control cluster variable x, respectively, and promote the normalized deviation of the corresponding variables to be less than 10% (average treatment effect). In this paper, closest-neighbor corresponding, radius matching, kernel corresponding, and caliper corresponding are mainly used to analyze the average processing outcome of matching results, as suggested by Baser [72]. If the matching results obtained by different matching methods are similar, the matching results are relatively robust [73].

3. Results

3.1. The Influence of Internet and Information Technology Usage Regarding the Choice of Fruit Farmer’s Sales Channels

From the model estimation outcomes in

Table 2. The impact of the use of internet information technology for selecting improved sales channels.

	Self-Sale	Middleman Sales	Cooperative Sales
	Coefficient (Standard Error)	Coefficient (Standard Error)	Coefficient (Standard Error)
lnx0 Use of Information Technology	0.016	0.051 ***	0.084 ***
	(0.016)	(0.017)	(0.017)
x1 gender	−0.027	−0.071	−0.006
	(0.059)	(0.062)	(0.063)
x2 age	−0.001	−0.005	−0.000
	(0.003)	(0.003)	(0.003)
x3 Educational level	0.012 *	−0.006	0.006
	(0.007)	(0.007)	(0.007)
x4 Planting scale	0.006	0.001	−0.000
	(0.013)	(0.014)	(0.014)
x5 Years of planting	0.002	−0.001	0.001
	(0.004)	(0.005)	(0.005)
x6 Whether to join a cooperative	0.249 ***	0.103 ***	0.158 ***
	(0.037)	(0.039)	(0.040)
x7 Degree of specialization	−0.001	1.045	0.530
	(0.647)	(0.678)	(0.687)
x8 Have you ever gone to work	0.039	−0.020	0.005
	(0.067)	(0.070)	(0.071)
x9 Ease of use of communication technology	−0.052 **	−0.022	−0.043
	(0.024)	(0.026)	(0.026)
x10 Planting Technology Training	−0.010	0.047	0.047
	(0.039)	(0.041)	(0.041)
x11 Soil fertility	−0.025	0.027	−0.023
	(0.028)	(0.029)	(0.029)
x12 Orchard fineness	0.028	0.014	0.005
	(0.041)	(0.043)	(0.044)
x13 How far the orchard is from the market	0.045	0.085 **	0.076 **
	(0.034)	(0.036)	(0.037)
_cons	0.416	−0.256	−0.282
	(0.595)	(0.624)	(0.631)
N	471	471	471
adj. R2	0.154	0.076	0.145

Note: * p < 0.10, ** p < 0.05, *** p < 0.01.

, it is indicated that the usage of IIT by fruit farmers mainly possesses a substantial influence on the two channels of intermediary sales and cooperative sales, but it has no significant impact on self-sale. The influence on the sales of middlemen and cooperatives is significantly positively correlated at the level of 1%, which indicates that the frequent use of IIT has increased the possibilities of fruit growers for selecting effective sales channels. The possible reason is that the apple, as a high-value and storage-resistant fruit, is more likely to attract middlemen and cooperatives for marketing and requires more market information. Compared to self-sale, the two rest are more important for IIT, and the use of IIT has a stronger dependency. In addition, the regression results also show that whether or not to join the cooperative and the distance of the orchard from the market also has a certain degree of positive impact on the choice of the above two channels, which further illustrates that high-value agricultural products have a strong effect on industrial organization and information tools.

The outcomes of using the OLS regression framework to estimate the sales channels of fruit farmers are presented in Table 2.

3.2. Estimation Results of the Heckman Model

In order to confirm the rationality of the estimated effects of the influence of IIT usage on the fruit grower’s income, this paper uses OLS and Heckman’s two-stage regression models to compare the projected results, as revealed in

Table 3. The influence of internet and information technology usage on measuring the fruit farmers’ family income.

	OLS	OLS	OLS	Heckman Agricultural Income		Heckman Supporting Income
	Total Revenue	Agricultural Income	Supporting Income	Income Equation	Choice Equation	Income Equation	Choice Equation
lnX0Use of Information Technology	0.156 *** (0.033)	0.134 *** (0.034)	0.085 ** (0.033)	0.123 *** (0.037)	−0.103 (0.055)	0.083 ** (0.032)	−0.035 (0.052)
x1 gender	0.073 (0.119)	0.099 (0.126)	0.039 (0.120)	0.110 (0.123)	0.148 (0.198)	0.041 (0.116)	0.037 (0.186)
x2 age	0.001 (0.006)	0.001 (0.006)	−0.001 (0.007)	−0.003 (0.007)	0.021 ** (0.011)	−0.001 (0.006)	0.005 (0.010)
x3 Educational level	0.015 (0.013)	0.012 (0.014)	−0.003 (0.013)	0.014 (0.013)	0.025 (0.027)	−0.001 (0.017)	0.021 (0.022)
x4 Planting scale	0.075 ** (0.027)	0.101 *** (0.028)	0.063 ** (0.034)	0.101 *** (0.027)	−0.012 (0.045)	0.064 ** (0.038)	0.003 (0.043)
x5 Planting scale	−0.005 (0.009)	−0.003 (0.009)	−0.004 (0.009)	−0.004 (0.009)	−0.013 (0.015)	−0.003 (0.009)	0.013 (0.014)
x6 Whether to join a cooperative	−0.044 (0.070)	0.165 (0.072)	−0.096 (0.069)	−0.147 ** (0.075)	0.350 (0.219)	−0.079 (0.070)	0.596 *** (0.209)
x7 specialization degree	0.464 (1.284)	0.743 (1.368)	0.180 (1.290)	1.190 (1.475)	4.417 *** (2.197)	0.223 (1.248)	0.510 (2.042)
x8 Have you ever gone to work	0.093 (0.135)	0.065 (0.141)	0.143 (0.137)	−0.092 (0.142)	0.225 (0.198)	0.073 (0.137)	0.271 (0.209)
x9 Ease of use of communication technology	−0.087 (0.051)	−0.062 (0.054)	−0.086 (0.051)	−0.036 (0.064)	0.247 *** (0.220)	−0.079 (0.050)	0.080 (0.076)
x10 Technical training	0.038 (0.079)	−0.008 (0.083)	0.037 (0.079)	−0.002 (0.081)	0.069 (0.125)	0.038 (0.076)	0.029 (0.121)
x11 Technical training	0.027 (0.054)	0.018 (0.057)	0.072 (0.055)	0.026 (0.056)	0.080 (0.089)	−0.057 (0.056)	0.202 (0.089)
x12 Orchard fineness	−0.109 (0.084)	−0.142 (0.008)	0.056 (0.084)	−0.158 (0.089)	−0.092 (0.143)	−0.072 (0.083)	−0.201 (0.128)
x13 How far the orchard is from the market	−0.028 (0.074)	−0.087 (0.079)	0.022 (0.080)	−0.124 (0.093)	0.317 (0.110)	−0.001 (0.077)	0.249 ** (0.106)
$W_i$	0.370 *** (0.108)	0.407 *** (0.126)	0.322 ** (0.123)	0.430 *** (0.126)	0.199 (0.205)	0.339 * (0.121)	0.172 (0.187)
$W_{i1}$	0.377 *** (0.134)	0.431 *** (0.141)	0.355 ** (0.139)	0.356 ** (0.158)	0.025 (0.192)	0.340 * (0.141)	0.092 (0.220)
$W_{i2}$	0.351 (0.345)	0.235 (0.418)	0.267 (0.378)	_	_	_	_
Constant term	9.694 *** (1.168)	8.366 ** (1.238)	9.960 *** (1.182)	8.399 *** (1.482)	−4.308 ** (2.073)	9.738 *** (1.175)	−1.309 (1.863)
Inverse Mills by	—	—	—	−0.243 *** (0.082)	—	−0.314 *** (0.063)	—
Prob > chi2	—	—	—	0.000	—	0.063	—
Prob > F	0.005	0.000	0.184	—	—	—	—

Note: * p < 0.10, ** p < 0.05, *** p < 0.01.

. The outcomes indicate that the principal explanatory variables, the use of internet information technology, OLS regression results, the p values passed the 1% or 5% significance test, and the inverse Mills ratio in the Heckman model passed the 1% significance assessment, indicating that the estimated model selection is appropriate.

Based on the results estimated by the OLS model, fruit farmers active usage of IIT has a positive impact on both their agricultural income and total income at a significant level of 1%, and at the same time it has a positive impact on their supporting income, which is also significant at the 5% level. In addition, the Heckman two-stage regression framework is utilized to remove the bias of fruit farmers’ participation in agricultural and supporting labor. The estimated results show that whether or not fruit farmers actively use IIT also has a substantial constructive influence on their agricultural and supporting income. However, the results obtained by using the OLS method to estimate the agricultural income are underestimated at 0.023 compared with the Heckman two-stage regression method for the supporting income. At the same time, the results obtained by using the OLS method are estimated at 0.017, which is lower than the Heckman two-stage regression method. In addition, the OLS method and Heckman’s two-stage regression assessment results show that the planting scale has an optimistic influence on agricultural, supporting, and overall income, and the assessment of the Heckman regression model is more persistent with the estimation outcomes of the OLS regression model. It specifies that with the increasing level of agricultural productivity, a moderate expansion of the orchard planting scale will help to further increase the income level of fruit farmers and also have an optimistic influence on the increase of supporting income. The possible reason is that the higher agricultural income of fruit farmers will foster the opportunity for deciding whether to engage in the supporting industry to obtain income. The results of group differences showed that the results of the OLS method and Heckman two-stage regression for fruit growers younger than 60 years old were consistent with those of the overall regression model, but the results of OLS method for fruit growers older than 60 years were not significant, and the results could not be obtained by the Heckman two-stage regression. The possible reason is that most farmers over the age of 60 stay at home, and it is difficult to obtain stable non-agricultural income by working outside. In general, the active use of IIT has a positive effect of more than 30% on the total income, agricultural income, and supporting the income of fruit farmers.

3.3. Matching Estimation of Fruit Farmers’ Internet Information Technology

Considering that the selection of matching factors must satisfy both the utilization of IIT and the fruit income of fruit farmers, this article selects the basic characteristics of the interviewees and the producer operating characteristics as matching variables and uses the OLS model to perform a logit regression estimation of propensity scores. It could be perceived from

Table 4. Logit regression estimation results of the propensity score of fruit farmers using internet information technology.

Indicator Name	Coefficient	Standard Deviation	p-Value	Indicator Name	Coefficient	Standard Deviation	p-Value
lnx0	0.247 ***	0.089	0.005	x7	5.495 *	(3.291)	0.095
x1	−0.274	0.306	0.371	x8	−0.078	(0.361)	0.830
x2	−0.025	0.016	0.121	x9	−0.117	(0.128)	0.363
x3	−0.032	0.035	0.364	x10	0.216	(0.199)	0.278
x4	−0.012	0.070	0.862	x11	0.118	(0.143)	0.408
x5	0.002	(0.023)	0.928	x12	0.048	(0.212)	0.823
x6	1.159 ***	(0.332)	0.000	x13	0.346 *	(0.189)	0.067
likelihood = −153.414	Pseudo R2 = 0.122	Prob > chi2 = 0.0001

Note: * p < 0.10, ** p < 0.05, *** p < 0.01.

that the use of internet information technology in deciding whether to join a cooperative, the degree of specialization of planting, and the distance of the orchard from the market have a significant positive impact on whether to actively use internet information technology, which is suitable for matching variables.

3.4. The Average Treatment Effect of Fruit Farmers’ Family Income

The estimated results regarding the influence of IIT on fruit farmers’ income are portrayed in

Table 5. The average processing effect of income by fruit farmers using internet information technology.

Dependent Variable	Matching Method	Processing Group/Control Group	Average Treatment Effect	Standard Deviation	t Value
Agricultural income	1:3 nearest neighbor matching	136/335	0.431 ***	0.159	2.72
	Radius matching (caliper 0.03)	136/335	0.367 ***	0.162	2.27
	K nearest neighbor matching in caliper (caliper 0.01)	136/335	0.414 ***	0.165	2.50
	Nuclear matching	136/335	0.315 ***	0.152	2.34
	ATT mean		0.392	—	—
Supporting income (Less than 60 years old)	1:3 nearest neighbor matching	136/335	0.329 **	0.163	2.32
	Radius matching (caliper 0.03)	136/335	0.325 **	0.165	2.26
	K nearest neighbor matching in caliper (caliper 0.01)	136/335	0.271 **	0.146	2.34
	Nuclear matching	136/335	0.312 **	0.158	2.35
	ATT mean		0.309	—	—
Supporting income (Farmers aged 60 and above)	1:3 nearest neighbor matching	136/335	0.172	0.332	0.52
	Radius matching (caliper 0.03)	136/335	0.265	0.456	0.58
	K nearest neighbor matching in caliper (caliper 0.01)	136/335	0.157	0.769	0.20
	Nuclear matching	136/335	0.129	0.415	0.31
	ATT mean		0.181	—	—
Total revenue	1:3 nearest neighbor matching	136/335	0.344 **	0.175	1.96
	Radius matching (caliper 0.03)	136/335	0.354 ***	0.161	2.20
	K nearest neighbor matching in caliper (caliper 0.01)	136/335	0.395 ***	0.174	2.27
	Nuclear matching	136/335	0.336 ***	0.151	2.22
	ATT mean		0.357	—	—

Note: * p < 0.10, ** p < 0.05, *** p < 0.01.

. The average treatment effect (ATT) of fruit farmers’ agricultural income obtained using the tendency matching method is significant at the level of 1%. The ATT results obtained using the four tendency matching methods are as follows: 0.431 (1:3 nearest neighbors matching), 0.367 (radius matching), 0.414 (k-nearest neighbor matching in calipers), and 0.355 (kernel matching). The difference in results obtained by the four matching methods is not large, which indicates that the matching results have a certain degree of stability. It also shows that after eliminating the apparent bias caused by observable heterogeneity, on average, the agricultural income is 39.2% higher for families that actively use IIT than for households in which IIT is not in active use (control group). From the results of using the four tendency matching methods to obtain the supporting income ATT(fruit farmers less than 60 years old), the average income treatment effect (ATT) is significant at the level of 10%, and the ATT results are 0.329 (1:3 nearest neighbors matching), 0.325 (Radius matching), 0.271 (k nearest neighbor matching in calipers), and 0.312 (nuclear matching). The results of the four matching methods are still close, indicating that the matching results are relatively stable. At the same time, after eliminating the apparent bias caused by observable heterogeneity, the average supporting income of active use of IIT is higher than the average supporting income of unused households, 30.9% higher. From the results of using the four tendency matching methods to obtain the supporting income ATT (fruit farmers aged 60 and above), the average income treatment effect (ATT) is not significant at the level of 10%. It shows that it is difficult for them to obtain stable non-agricultural income by working outside and farming at home may be the best choice. Finally, from the estimation results that affect the total income of fruit farmers, the average treatment effect (ATT) of the total income of fruit farmers using the four tendency matching methods is significant at the level of 1%, and the ATT results are: 0.344 (1:3 nearest neighbors) matching), 0.354 (radius matching), 0.395 (k nearest neighbor matching in calipers), and 0.336 (kernel matching). The results of the four matching methods are still relatively close and further illustrate the stability of the matching results. It shows that after eliminating the apparent bias caused by observable heterogeneity, the total income of fruit farmers who actively use IIT is 35.7% higher than that of fruit farmers who are not actively using it.

3.5. Balance Analysis of Matching Variables and Comparative Analysis of Different Estimation Methods

According to the research conclusions of Rosenbaum & Rubin [71], only when there is no substantial change in the matching constructs before and after the matching, then the tendency matching results are more reliable and convincing. While the larger the p-value of the matching variable after matching, the better the matching effect and the more reliable the tendency matching estimation.

Table 6. Parallel hypothesis test before and after propensity matching.

Variable Name	Before and After Matching	Processing Group	Control Group	Difference Rate%	Change Rate %	p-Value
lnx0 Internet information Use of technology	Before matching	2.261	1.250	48.48		0.001
	After matching	2.137	2.036	4.9	90.0	0.913
x6 Whether to join cooperative	Before matching	0.916	0.609	44.8		0.003
	After matching	0.851	0.804	6.9	84.5	0.671
x7 specialization degree	Before matching	0.777	0.770	16.2		0.199
	After matching	0.776	0.778	−4.4	73.0	0.685
x13 Orchard off the market Distance	Before matching	1.534	21.94	31.8		0.008
	After matching	1.464	1.294	−25.2	20.7	0.153

portraits the comparison of the treatment cluster and the control cluster based on the mean values of the matched factors before and after the coordinated factors match. As a whole, there are major alterations in the four matching variables, the utilization of IIT, whether to join a cooperative, the degree of specialization in planting, and the distance of the orchard from the market before the match, and the p-value of each variable significantly increases after matching. However, the test results are not significant, indicating that whether to actively use IIT after matching is not affected by its basic characteristic variables, the matching effect is good, and the estimation result is more reliable. In addition, the effects of IIT use by farmers’ incomes obtained from different methods can be seen from the comparison (see

Table 7. Comparative analysis of regression results of internet information technology use on fruit farmers’ income differences.

Regression Results	OLS Returns	Heckman Returns	PSM Regression Mean	Deviation from PSM Selection
Agricultural income	0.407	0.430	0.392	0.015\0.038
Supporting income ($W_i$)	0.322	0.339	0.366	−0.044\−0.027
Supporting income ($W_{i1}$)	0.377	0.340	0.309	−0.068\−0.031
Supporting income ($W_{i2}$)	0.351	_	0.181	Not significant
Total revenue	0.370	——	0.357	0.013

). Although the results are slightly different, the overall deviation is not large. Specifically, compared with the results of the propensity score matching, the regression results of OLS estimation overestimated the total income increase of farmers by 1.3%. From the perspective of agricultural income comparison, the regression results of OLS estimates overestimated the increase of agricultural income by 1.5%. The estimated result of the Heckman regression model overestimates the effect of increasing agricultural income by 3.8%. From the comparison of supporting income, compared with the tendency score matching estimation result, the income increasing effect of the OLS estimation regression result is underestimated by 4.4%. The supporting income increase effect is underestimated by 2.7%. The results of differences among groups show that the OLS estimation of fruit farmers younger than 60 underestimates the effect of the income increase by 6.8%, and the Heckman regression model underestimates the effect of the non-agricultural income increase by 3.1%. In addition, the results of three regression methods for fruit farmers older than 60 are not significant. It further shows that most farmers over the age of 60 find it difficult to obtain stable non-agricultural income by working outside. In theory, the different methods are not comparable, but using the Heckman two-stage regression and the PSM propensity score matching tactics to some extent remove the discernment bias generated by the OLS estimation and tests the robustness of each other [74]. This demonstrates that the usage of IIT possesses a significant impact on the incomes of fruit farmers’ families, which is parallel with the findings of Rolfe et al. [75], Ankrah Twumasi et al. [76], Gloy and Akridge [77], and other recent studies. On average, the effects of active use of IIT on agricultural, supporting, and overall income of fruit farmers’ are exceeded by 30%. Among them, the most prominent effect is agricultural income, with an income increase effect of about 40%, followed by non-agricultural income and total income.

4. Conclusions

The agribusiness industries in rising economies are currently facing severe challenges from mitigating the ever-increasing production demand via sustainable intensifications to the development of job opportunities for vulnerable rural communities. Additionally, worldwide economic development trends and the rapid expansion of structural changes have a staggering effect towards this fundamental sector. Such characteristics craft an outstanding demand, as the dynamics of IIT could be utilized to deal with those potential issues and cope with the advancement of the global economy while maintaining a sound transition of production and distributions. IIT possesses pro-vital impacts by empowering farmers to facilitate smooth access towards potential market information, while covering limitless agroecological scenarios. Thus, it fosters an ample interest within the industry practitioners, academia, policy crafters, and international communities regarding farm-level interactions of IIT. The study tends to explore whether the interaction of internet and information technology (IIT) can foster the betterment of choosing the right sales and distribution network and eventually help farmers to improve their agricultural, nonagricultural, and overall income. Interestingly, the existing research has mainly compiled the impacts of information technology and internet within a separate dimension, while we have explored the internet and information technology as an integrated manner. Moreover, compared with the existing literature, we chose a high value agricultural product like apple rather than grain, which may possess higher research values. The empirical setup of the study has been comprised by 471 apple farmers within nine counties (cities, districts) of Shandong Province. More specifically, we have utilized OLS regression, Heckman two-stage regression, and propensity score matching (PSM) for crafting the assessment. The empirical evaluation demonstrated that the higher the level of IIT interaction possessed by the famers the greater the possibilities to capture an effective sales channel while the impact of self-sales were found to be relatively lower. The adopted model assessment also revealed that the active use of IIT has increased the total income of fruit farmers, agricultural income, and supporting income by more than 35%. While the impact of IIT on agricultural income increase is found to be the most prominent, the average income increase effect is (about 41.1%), followed by non-agricultural income (about 32.45%). As a high-value agricultural product, with apple it is easier to get a higher agricultural income than other agricultural products, so it will have a higher opportunity cost for off-farm work. The assessment also found similar results, as it indicated a lower incensement of supporting or non-agricultural income. Interestingly, the regression estimation result of different age groups shows that fruit farmers over 60 years are more suited to get more agricultural income through apple planting while the non-agricultural income is not significant for them. Nonagricultural income mainly comes from young and middle-aged fruit farmers under 60 years old. Based on the above conclusions and discussion, the study outlines the following. First of all, the government should implement diversified training facilities and broaden the viability of sales and distributions channels for the smooth dissemination of basic information regarding sales and marketing. Public and private investment should be facilitated while an agricultural demonstration zone also needs to be used for capturing valuable information in rural areas. Awareness building campaigns should also be conducted to remove the adoptability issues of basic IIT especially focusing on the older generation of farmers. However, the public and private sector partnership should be promoted for fostering the development of agriproduct information sharing platforms, increasing the coverage of promotional and marketing facilities, and delivering valuable agricultural innovation to the majority of fruit farmers in a timely manner. To foster competitiveness, several distinct categories of marketing and distributional entities (such as online and apps based on the agriproduct selling marketplace) should be encouraged to assist in the selling structure of farm products and to develop a structured mechanism for sound connection among farmers, middlemen, and collaborative networks.

While the research examines the importance of IIT for the effective management of available distributional networks among apple farmers, there are still some limitations that should be studied further by the future researcher. First of all, as Shandong is plain land, there could be the possibility that middleman and other distributional agents often visit the farms to buy apples, which could have had some impact on improving income as well. Thus, future research should include this impact while constructing the model. In reality, in an infrastructure centered on constant development and fundamental innovation, the alternative may be valid. Potential studies should explore how and to what extent the IIT contributes to the farmer’s market involvement for facilitating transition costs within the context of supply chain resources and economic efficiency. The index design of non-agricultural employment has not been deeply evaluated, rather, to be able to craft the non-agricultural employment issue, we have explored the frequency of migrant work through age characteristics to indirectly indicate whether the fruit farmers are out of town for work. Potential research should explore the issue more deeply, and if the actual transitional effects could be traced, it will be more interesting. Secondly, the level of economic development is different in different regions. Therefore, the off-farm employment might be different. Moreover, in some areas, local farmers can achieve non-agricultural employment within their territories without going outside to pursue employment. This may lead to some errors in experimental results. In the future, potential studies should divide the types of employment of farmers from the perspective of the industrial economy by refining the design of the questionnaire, adding variables to reflect non-agricultural employment, and refining the research content of non-agricultural employment.