The role of specialization in increasing farmers' income: Horizontal or vertical?

Based on the theory of division of labor and specialization, this article attempts to address the problem of “splitting horizontal and vertical division of labor” by innovatively integrating horizontal specialization, vertical specialization, and farmers' income into a unified analytical framework. Using the data for 416 chicken farmers in four districts and counties of Guangdong Province as the sample, the theoretical framework proposed in this article is tested by constructing multiple linear regression models and structural equation models. The results are as follows: Both vertical and horizontal specialization can bolster the income of farmers, but they have different specific mechanisms of action. Vertical specialization promotes agricultural income through the mediating effect of horizontal specialization. It is evident that the discussion of horizontal specialization of farmers should not be separated from the conditions of vertical specialization. The policy effect of emphasizing horizontal specialization alone may be limited. Especially for those farmers who are limited by their own endowments and objective conditions to increase their horizontal specialization level and production scale in the short term, it is more important to promote their vertical specialization level.


INTRODUCTION
The specialized production mode is a fundamental way to improve production efficiency and producer income, [1][2][3][4] which seems to be noncontroversial and the oldest theory in economics.The same argument applies to the agricultural sector.][7][8][9] Some scholars further divide specialization into two dimensions, horizontal and vertical, to discuss the effects of specialization on increases in farmers' income.Zou pointed out that horizontal specialization is a process in which farmers specialize in the production of a single variety and increase the concentration of production, which is conducive to realizing economies of scale and endogenous specialized economies and improving comparative agricultural benefits. 10Based on transaction cost theory, Omamo also concludes that horizontal specialization helps improve farmers' income. 11Hu delineated that vertical specialization or division of labor permits farmers to streamline operational procedures through socialized services and concentrate on a handful of production stages.For instance, grain growers can separate various processes, such as soil preparation, sowing, fertilization, pesticide application, and harvesting, by purchasing these mechanical services from societal entities, thereby specializing solely in daily care activities.Similarly, poultry farmers delegate tasks like incubation, daily disease prevention, transportation, and even the sales of chickens to external parties providing social services, focusing only on feeding.This division of labor allows for the realization of "service economies of scale," which can augment income levels. 12][15] Although the studies literature has respectively discussed the effect of horizontal specialization and vertical specialization on farmers' income, in the mainstream economics literature, the theory of horizontal division of labor characterized by horizontal specialization has been relatively separate from the theory of vertical division of labor characterized by vertical specialization. 16In recent years, some scholars have begun to explore the relationship between horizontal specialization and vertical specialization in agriculture.Luo argues that it is difficult to deepen the vertical division of labor in which farmers work without a horizontal division of labor, meaning horizontal specialization is a prerequisite for vertical specialization. 16Another voice pointed out that the deepening of the vertical division of labor can lead to the involvement of farmers in horizontal specialization across a larger area; that is, vertical specialization promotes horizontal specialization. 17n summary, there are still no consensus in the literature on the relationship between horizontal specialization and vertical specialization.In addition, when the existing studies discuss the relationship between specialization and farmers' income, horizontal specialization and vertical specialization are mostly considered separately, and they do not further incorporate horizontal specialization, vertical specialization, and farmers' income into a unified analytical framework.The current study is based on the theory of division of labor and specialization but attempts to avoid the separation of horizontal and vertical division of labor by innovatively incorporating horizontal specialization, vertical specialization, and farmers' income into a unified analytical framework.We conducted a questionnaire survey of 416 chicken farmers in four districts (Kaiping, Xinxing, Gaoyao, and Zengcheng) of Guangdong Province, China.To the data obtained, we applied multiple linear regression models, structural equation models, and other empirical methods to analyze the differences in the effects of different specialization dimensions on farmer's income.The contribution made by this article is to prove that both vertical specialization and horizontal specialization can bolster the income of farmers, but the specific mechanisms of action between the two are different, and the specific manifestation is that vertical specialization can boost farmers' income through horizontal specialization, that is, there is a partial mediating effect.The structure of this article is as follows: The Section 2 discusses the relevant specialization theories and proposes research hypotheses based on them; the Section 3 clarifies the data sources and the relevant variables and the selection of the models; the Section 4 presents the results of the empirical analysis and the support they give to the research hypotheses; and the Section 5 summarizes the empirical analysis and proposes relevant policy implications.

THEORY AND HYPOTHESIS
Compared with nonspecialized production, specialization has a lower average cost and higher output per labor.This argument has been a consensus in the field of economics since Adam Smith.The realization of economies of scale lies in the fact that specialized production enables a single producer to reduce the average cost of products with the expansion of production scale under the premise that other factors remain the same, thereby realizing economies of scale.Therefore, economies of scale originate from the division of labor and specialization. 18ue to the different dimensions of the division of labor, specialization can be further subdivided into horizontal specialization and vertical specialization.The similarities between horizontal and vertical specialization in achieving economies of scale are (1) they both reduce operational processes and improve efficiency; (2) they both reduce the time wasted by the producer in the conversion between different operations and save human resources; (3) they both reduce the variety of production materials and capital occupation; (4) they both simplify the production technology; and (5) they both promote the development of roundabout methods of production. 19In agriculture, the difference between two lies in the following: vertical specialization reduces the operational processes of farmers through socialized services and makes them specialize in a small number of production processes, corresponding to a large-scale service operation.Horizontal specialization enables farmers to specialize in the production of a single species and expand their scale of production, which corresponds to large-scale land management. 16he realization of economies of scale in agricultural production mainly comes from both horizontal specialization and vertical specialization: 1. Expanding the production scale is inseparable from horizontal specialization.Due to the rigid constraints of the labor force, the larger the production scale of one agricultural product by a single farmer, the less room the farmer has to engage in part-time business or in the production of other agricultural products, and the higher the level of horizontal specialization.Therefore, small-scale production and horizontal specialization are contradictory, and horizontal specialization means a larger production scale.Since specialized production can achieve higher production efficiency, the expansion of the production scale for a single product and the increase in the horizontal specialization level can enable farmers to achieve economies of scale.The realization of economies of scale means that their profits and income level increase. 1,18On this basis, a hypothesis (H1) is proposed: H1: The level of horizontal specialization of farmers has a positive impact on their agricultural income.
2. In agricultural production, vertical specialization is primarily exemplified through the utilization of indirect and socialized production methods.This involves purchasing productive services from societal entities such as "company plus farmer" models, cooperatives, and service outsourcing. 20,21When the cost of personal production for certain processes surpasses the cost of procuring intermediate products or services from the market, it is more economical to delegate these production tasks to the market to curtail costs and augment profits. 22Empirical studies corroborate this approach, demonstrating that the lower the price of acquisition and utilization of agricultural productive services (such as mechanical services), the farmers more tend to purchase services, which can effectively enhance the income levels of individual farmers. 23Service providers cater to a multitude of farmers, resulting in a lower equipment vacancy rate, operational proficiency in their personnel, and a high degree of specialization, thereby yielding a product average cost that is lower than that of a single producer.Additionally, a reduction in the operational process can decrease the steps in which they must invest and increase farmers' production efficiency in a few production stages.As such, farmers achieve economies of scale through vertical specialization, leading to higher income levels.Based on this, we propose a second hypothesis: H2: The level of vertical specialization of farmers has a positive impact on their agricultural income.
Let us delve further into the relationship between vertical and horizontal specialization.If transaction efficiency is high, individuals can opt for an elevated level of vertical division of labor while sustaining the high level of specialization of each entity through the division of labor among numerous specialized producers. 4Consequently, through vertical division of labor, the reduction in operational steps, and the deployment of vertical specialization production methodology, each producer can focus on a limited varieties of products, thus achieving horizontal specialization.Additionally, the division of labor and specialized production yield higher efficiency compared to self-sufficiency.When choosing specialized products, individual producers will gravitate towards those with comparative advantages. 24elevant research within the agricultural sector supports this assertion.For instance, Takeshima empirical research on Nigerian farmers demonstrates that the procurement of mechanical services (thus enhancing the level of vertical specialization) enables farmers to concentrate on fewer varieties of agricultural produce, resulting in the realization of horizontal specialization. 25This could be due to the fact that adopting vertical specialization can reduce farmers' operational processes, saving both time and energy.Furthermore, given their fewer operational steps, farmers can invest less in such processes, which significantly reduces the asset specificity of agricultural production.
On the one side, purchasing services to reduce operational procedures can objectively enable farmers to spend their limited time, energy, and resources on a single operational link to expand production, thus improving their horizontal specialization level.To be specific, because the labor force and resources of individual farmers are very limited, it is difficult for them to expand the scale of production and improve the level of horizontal specialization without changing their labor input, nor can they achieve vertical division of labor internally.At this time, farmers can only achieve vertical division of labor by purchasing services from the society, thus improving the production scale and horizontal specialization level.Therefore, farmers will expand their production scale and improve their horizontal specialization level only on the premise that vertical specialization has reached or is expected to reach a certain level.On the other side, the adoption of vertical specialization production methodology provides farmers with a comparative advantage in a few production links.Subjectively, this makes them more inclined to expand their production scale in a few production links and enhance their level of horizontal specialization.Consequently, in agricultural production, vertical specialization plays a pivotal role in facilitating the achievement of horizontal specialization.On this basis, a third hypothesis is proposed: H3: The vertical specialization level of farmers has a positive impact on their horizontal specialization level.
Based on the analysis of the relationship between the horizontal specialization level, the vertical specialization level and the income level of the farmers in hypotheses H1, H2, and H3, it can be seen that both horizontal specialization and vertical specialization promote the agricultural income growth of the farmers.Moreover, the level of vertical specialization can have a positive impact on the level of horizontal specialization.Therefore, it can be further deduced that vertical specialization may have an impact on the agricultural income of farmers through horizontal specialization.On this basis, a fourth hypothesis is proposed: H4: The horizontal specialization level plays a mediating role in the impact of the vertical specialization level on the agricultural income of farmers.

Data source
In this study, chicken farmers in Kaiping, Zengcheng, Gaoyao, and Xinxing districts were selected as the study subjects for the following reasons: (1) To exclude the influence of geographical factors on the survey results, it may be more appropriate to choose chicken farmers than to choose growers.Because chicken farmers are less affected by land leveling conditions and chicken farming is a relatively common agricultural industry in South China, the impact of local product specificity (specialty) is excluded.(2) The chicken farmers in the four districts/counties have relatively diverse organizational forms, including Wen's organizational forms of large companies plus farmers and small companies plus farmers, with dozens of household scales, as well as the single-farmer form.The survey subjects have distinct organization levels and substantial differences between them.In this study, a total of 600 questionnaires (predistribution and formal distribution) were distributed in the above four districts and counties, and 416 valid questionnaires were recovered.Among them, 100 copies were obtained by the members of the research group through household visits, on-site explanations, on-site filling, and on-site receiving for representative chicken farmers in accordance with the recommendations of the local poultry industry association, which were well representative.This method of receiving copies can be up to 100% efficient.Another 500 copies were distributed on behalf of the farmers through the technical service personnel of the breeding company during the inspection of the farmers, because the technical service personnel of each company generally conduct inspections of their farmers in accordance with the established order (in accordance with the principle of "following the road").Therefore, the practice of the technical service personnel of each company inspecting the farmers can be regarded as random sampling with strong randomness.In this manner, 316 valid questionnaires were recovered, and the recovery efficiency was 63%.

3.2
Variable construction and descriptive statistics

Explained variable: Income level
If the proportion of the increase in the labor factor input is smaller than the proportion of the increase in output, this is a concrete manifestation of economy of scale, as reflected in the increase in the per capita income of the producers.In this study, the per capita annual income of farming household can be expressed as the following formula: where I is the per capita annual income of the producer, P is the average profit per chicken of the chicken farmer, S is the scale of operation of the chicken farmer, G is the number of batches of chickens that are ready for the market each year, and N is the number of workers on the chicken farm.

Mediating variable: Horizontal specialization level
Horizontal specialization in farming means reducing the number of varieties produced and expanding the production scale of a single or a few varieties. 16Most studies use the average of the three indicators of "production type concentration," "production scale concentration," and "sales scale concentration" to determine the horizontal specialization level of farmers. 10Although this method can comprehensively reflect the horizontal specialization level of farmers, there are many problems involved in the questionnaire, and the complexity of the questionnaire is relatively high.Moreover, taking the average of the three indicators does not consider differences in their importance, that is, the issue of weight.
Given the needs of this research, this article optimizes the above indicators based on the idea of "concentration" and uses the per capita breeding scale to determine the horizontal specialization level of chicken farmers.The reason for this approach is that the survey area mostly adopts free-range chicken farming, which is different from cage farming.In the production process, most of the steps cannot use machinery to replace human labor, so the chicken farming industry in the survey area requires a high labor factor input.Due to the rigid constraints of the labor force (time and energy), the production activities that a labor force can support are very limited.Therefore, the larger the per capita production scale is, the less likely it is that the producer will engage in part-time business or produce other agricultural products, which means higher levels of production, sales concentration and horizontal specialization.Therefore, the use of the per capita farming scale as a measure of the horizontal specialization level in this study is reasonable.It is calculated as where H is the horizontal specialization level of the chicken farmers, S is the operating scale of the farmer household, and N is the number of workers in the farmer household.

Core explanatory variable: Vertical specialization level
Vertical specialization corresponds to the vertical division of labor, that is, a single producer specializes in a specific stage of producing a commodity. 26,27In agricultural production, vertical specialization is reflected in the division of labor and specialization in the production process of the same agricultural product, that is, different production links are completed by different producers. 12This form of specialization and division of labor is realized mainly through socialization and roundabout production methods, which is embodied in productive service purchasing. 14,20,21his holds true in the present study as well.The majority of the respondents are small farms operating on a family basis, often resembling "mom-and-pop" establishments.Under this organizational model, the realization of vertical division of labor internally is challenging due to limited personnel.As a result, the subject of this research can only attain vertical specialization through societal vertical division of labor (by purchasing services from society).Consequently, the degree of farmer productive service purchasing is directly proportional to the level of vertical specialization.This article plans to adopt a prevalent method to measure the vertical specialization level of poultry farmers from the ratio of actual expenditure on productive service purchases to the expenditure required for all purchasable components. 28It is calculated as follows: In the above formula, L denotes the vertical specialization level of the farmers, R represents the actual household expenditure for service purchases, and W is the expenditure necessary for all purchasable components.

Control variables
Considering the existing empirical research in the field of farmer specialization, 5,7,9,29 along with the characteristics of the survey and the sample data in this study, as well as the fact that age, gender, education level, native, 30 amount of family-owned arable land, amount of family-owned forestland, household size, number of migrant laborers in the household, and opportunity cost (the income of migrant workers forgone by the farmer to engage in the chicken farming industry) may have an impact on the income level, these variables are set as the control variables.

Descriptive statistics
Descriptive statistics are performed on the variable data, and the results are shown in Table 1.Table 1 reveals the following: 1.The average income level of the sample is 50,437.229yuan, and the standard deviation is 30,301.445yuan.The sample average has exceeded the average wage level of migrant workers in the country in 2020 (48,864 yuan). 31Thus, specialized agricultural production can bring richer returns and has advantages over other industries.2. The average horizontal specialization level of the sample is 6626, the standard deviation is 4625, and the maximum value is 40,000, indicating that the horizontal specialization level of most respondents must be improved.
3. The sample vertical specialization level has a mean of 0.669 and a standard deviation of 0.281.These figures show that the socialized production methods of the survey respondents are insufficiently used, some purchasable links have not been purchased, and the level of vertical specialization must be improved.4. The average age of the surveyed subjects was 49.712 years, and the oldest was 69 years old, indicating that the survey subjects were fairly old. 5.The average gender of the sample is 0.793, indicating that men are dominant in the chicken breeding industry under investigation.6.The average education level of the farmer was 1.933, indicating that the farmer's education level was mostly up to elementary or middle school, for a relatively low level of education.Given their schooling and age, chicken farming at home may be a better choice than to work outside.7. The respondents were mostly from native, but 23.3% of nonlocal people engaged in chicken farming.The main reason for this finding lies in the relatively high degree of agglomeration and division of labor in the production chain of the surveyed chicken breeding industry and the low cost of using social services.8.The surveyed farmers generally own less arable land and more forestland, which is determined by the geographical features of the surveyed area with many hills.Sloping fields are exactly what chicken farmers need.9.The average population of the sampled households was 5.120, and the average number of migrant labors was 1.421, indicating that migrant workers are more common in the local area.Moreover, according to the sample, the average monthly income (opportunity cost) of migrant workers is estimated to reach 4279.808yuan, indicating that local migrant workers' income is generally high, which reduces the objective attractiveness of agriculture.

Measurement method selection
Since this study uses horizontal specialization as the mediating variable of the relationship between vertical specialization level and farmers' income level, the mediating effect is tested in three steps as follows: 32 (1) Model 1 is established to test the influence of vertical specialization level on income level; see Equation ( 4). ( 2) Model 2 is established to test the influence of the vertical specialization level on the horizontal specialization level; see Equation (5).(3) Model 3 is established to examine the impact of vertical and horizontal specialization levels on income levels, as shown in Equation (6).Since the dependent variables of the above three models are all continuous variables, multiple linear regression models are all applicable.
Accordingly, the robustness of the results of the multiple linear regression model is tested with a structural equation model.

Measurement method selection: Results of the multiple linear regression model
This study uses STATA 13.0 and uses a multiple linear regression model to analyze the influence of the horizontal and vertical specialization level on chicken farmers' income level.The empirical test results are shown in Table 2. On this basis, the variance inflation factor (VIF) test was performed, and the test results showed that the VIF values of all independent variables were less than 10, and the average VIF = 1.29, indicating that there was no obvious multicollinearity problem among the independent variables.

Analysis of the results
First, Model 1 in Table 2 shows that the horizontal specialization level of chicken farmers has a significant positive impact on their income level, and thus, hypothesis H1 is supported.With the expansion of the production scale of chicken farmers, the proportion of labor input at a decreasing rate, that is, the per capita production scale and the level of horizontal specialization increase with the expansion of the breeding scale.The expansion of per capita production scale means improved production efficiency, which in turn increases the level of per capita output and income.Second, Model 3 in Table 2 shows that the vertical specialization level of chicken farmers has a significant positive impact on their income level, and thus, hypothesis H2 is supported.Chicken farmers with a high level of vertical specialization often focus only on the daily management link, while other production links are completed by service purchasing, and the average cost is reduced through "service economies of scale".For example, in the purchase of feed, veterinary drugs, and transportation services, agricultural social service organizations have stronger bargaining power than retail investors, so they can provide agricultural materials and services to chicken farmers at lower prices.Regarding the sales link, agricultural social service organizations have stable and convenient sales channels, and their sales speed is thus often faster than that of retail investors.Faster clearance speed means lower breeding costs.The reduction of such expenses will reduce the average cost of chicken farmers and increase their income level.
Third, Model 2 in Table 2 shows that the vertical specialization level of chicken farmers has a significant positive impact on their horizontal specialization level, and thus, hypothesis H3 is supported.Chicken farmers with a higher level of vertical specialization tend to have higher per capita production scale and higher level of horizontal specialization.The reason for this finding is that stocking methods are generally adopted in the research area, the mechanical substitution rate of each production link is very low, and the labor factor imposes strict constraints on the scale of breeding.Under this condition, if chicken farmers want to support a larger breeding scale, they must strip off more production links through service purchases, thus promoting horizontal specialization through vertical specialization.
Fourth, because hypotheses H1, H2, and H3 are supported, the horizontal specialization level of farmers seems to play a partial mediating role in the relationship between the vertical specialization level and the farming income, thus supporting hypothesis H4.

Robustness test
The robustness of the study results was tested.According to the actual needs, robustness is tested by changing the measurement method.Based on the logic that the horizontal and vertical specialization levels of farmers have a significant positive impact on their farming income, and the horizontal specialization level plays a mediating role between the vertical specialization level and the farming income, the structural equation model was established through AMOS 17.0 software to test the robustness of the above results.The structural equation model is shown in Figure 1.
The results of the derived structural equations are shown in Table 4.The horizontal and vertical specialization levels of farming households have a significant positive impact on their farming income (Table 4).In addition, the horizontal specialization level of farmers plays a mediating role in the impact of vertical specialization on farming income.There is no difference between this conclusion and the conclusion drawn from Table 2, indicating that the conclusions drawn from Table 2 are relatively robust.

CONCLUSIONS AND IMPLICATIONS
Based on the theory of division of labor and specialization, this study attempts to avoid considering the horizontal division of labor and vertical division of labor separately; we innovatively incorporate horizontal specialization, vertical specialization, and farmers' income into a unified analytical framework; and we analyze the difference in the contributions of different specialization dimensions to the increase in farmers' income.The proposed hypotheses were tested by constructing multiple linear regression models and structural equation models from data gathered from a questionnaire survey of 416 chicken farmers in Kaiping, Xinxing, Gaoyao, and Zengcheng, Guangdong Province, China.The results show that the impact of horizontal specialization on the agricultural income of farmers is different from that of vertical specialization.
Increasing the levels of horizontal and vertical specialization can improve production efficiency, thereby increasing the agricultural income.On the other hand, greater vertical specialization lets farmers reduce the number of production processes they engage in and spend their limited time, energy, and resources on the expansion of the scale of production in a few production steps, thereby increasing their horizontal specialization and increasing their agricultural income.That is, the horizontal specialization level plays a mediating role in the impact of the vertical specialization level on agricultural income.Based on the above research conclusions, this article proposes that discussion of the horizontal specialization of farmers should not be separated from discussion of their vertical specialization.Therefore, for farmers who are limited by their own endowments and objective conditions and cannot easily improve their horizontal specialization level and production scale in the short term, the ability of policy to continuing trying to promote the use of horizontal specialization production methods is limited.Policies should employ tax incentives and subsidies to attract agricultural social service providers (such as "companies + farmers" models and cooperatives) into agricultural communities.This can promote the amalgamation of farmers with social service organizations, making it more convenient for farmers to purchase services from society.Furthermore, policy should encourage manufacturers and research institutions to innovate and disseminate modern agricultural technology, thereby reducing the cost and price of service provision.This can potentially attract farmers to increase their utilization of social services, subsequently enhancing the level of vertical specialization and agricultural income.It also creates conditions for farmers to expand their production scale and level of horizontal specialization through improvements in vertical specialization.

CONFLICT OF INTEREST STATEMENT
Authors have no conflict of interest relevant to this article.

F I G U R E 1
Structural equation model of the impacts of horizontal and vertical specialization on farming income.
Descriptive statistics of the control variables TA B L E 1 Mediating effect of the horizontal specialization level of the chicken farmers between their vertical specialization level and TA B L E 2 Overall fit index of the structural equation model.