Sustainable Agriculture and Rural Poverty Eradication in Pakistan: The Role of Foreign Aid and Government Policies

: For decades, agriculture has been central to economic growth and development in Pak-istan. However, endemic rural poverty hinders the performance of agricultural production


Introduction
The 17 Sustainable Development Goals (SDGs) outlined by the economic and social affairs department of the United Nations were considered the areas to be dealt with urgently, and are known as the "2030 agenda" for sustainable development. The first three SDGs are (1) no poverty, (2) zero hunger, and (3) good health and well-being. These three SDGs goals are directly associated with the quality of sustainable agricultural production. Government departments, financial and non-financial sectors, and non-profit organizations in developing or under-developing economies need extra financial support or aid to achieve these SDGs. However, due to budgetary issues or funds, governments in developing countries cannot provide policy guidelines to relevant sectors. For instance, Pakistan's government has not changed the policies and strategies for the agriculture sector in the last decade. Hence, the agriculture sector fails to produce the required level of output. In such conditions, attracting foreign aid and allocating it to the most vulnerable sectors through

Selection of Variables
The dependent variable in this study is poverty reduction, which is defined as the economic growth which permanently lifts as many people as possible out of poverty (above the poverty line) in the short run ( Figure 1) [18]. Poverty reduction is measured based on the scale adapted from Okibo and Makanga [19].
Independent variables include sustainable agriculture, agricultural production, government policies, and foreign aid as moderating factors. Sustainable agriculture is defined as production to meet the demand of the present without compromising the needs of future generations [20][21][22][23]. It cannot be achieved without the cooperation of private and public sectors based on sustainable standards [24]. Following Allahyari et al. [25], sustainable agriculture is measured as a multidimensional construct based on social responsibility, environmental sustainability, economic viability, and production efficiency. Agricultural production means achieving sustainable outcomes through different state, market, and civil society interventions [26]. This study measures it based on the scale adapted from Trojan [27]. The measures taken by the government include providing incentives to producers, introducing more stringent standards, and facilitating the development of downstream Sustainability 2022, 14,14751 3 of 14 capabilities in order to achieve the targeted outputs [28] measured based on the scale derived from Deng et al. [29]. Independent variables include sustainable agriculture, agricultural production, government policies, and foreign aid as moderating factors. Sustainable agriculture is defined as production to meet the demand of the present without compromising the needs of future generations [20][21][22][23]. It cannot be achieved without the cooperation of private and public sectors based on sustainable standards [24]. Following Allahyari et al. [25], sustainable agriculture is measured as a multidimensional construct based on social responsibility, environmental sustainability, economic viability, and production efficiency. Agricultural production means achieving sustainable outcomes through different state, market, and civil society interventions [26]. This study measures it based on the scale adapted from Trojan [27]. The measures taken by the government include providing incentives to producers, introducing more stringent standards, and facilitating the development of downstream capabilities in order to achieve the targeted outputs [28] measured based on the scale derived from Deng et al. [29].

Research Approach
According to the United Nations Development Programme (UNDP), 54.6% of people in the rural area of Pakistan face poverty, which significantly influences their education, living standards, and leisure activities [30,31]. The province of Punjab in Pakistan is the second largest province in terms of rural population and the first in agriculture production. Therefore, it is considered one of the most affected territories by multidimensional poverty [30]. The irrigation system of Punjab is extensive, and is located in the semi-arid plains zone [32]. With 56.2% arable land, Punjab contributes 53% of the total agricultural contribution to the GDP, and 74% of cereal output [33]. Despite this magnificent infrastructure, the contribution towards the agriculture employment generation is not considered at all; thus, 31% of the rural people of Punjab live below the poverty line [34]. The total population of Punjab is 110 million, while that of Southern Punjab is about 34.74 million, with an average of five household members [35]. Southern Punjab is one of the country's leading territories in agriculture output, and despite that fact, most of the population lives below the poverty line. Hence, the present study selects Southern Punjab as the population for evaluation of the factors influencing poverty. The recent report published by the Government of Punjab indicates that Dera Ghazi Khan, Faisalabad, Gujranwala, Multan, Sahiwal, and Sargodha are the primary divisions in terms of maize, rice, wheat, cotton, and sugarcane production. The present study considered these divisions and associated tehsils as the population reported in Appendix A Table A6.
The present study used the nonprobability sampling technique in this research to collect data. It is considered to be convenient and suitable for this research. This technique

Research Approach
According to the United Nations Development Programme (UNDP), 54.6% of people in the rural area of Pakistan face poverty, which significantly influences their education, living standards, and leisure activities [30,31]. The province of Punjab in Pakistan is the second largest province in terms of rural population and the first in agriculture production. Therefore, it is considered one of the most affected territories by multidimensional poverty [30]. The irrigation system of Punjab is extensive, and is located in the semi-arid plains zone [32]. With 56.2% arable land, Punjab contributes 53% of the total agricultural contribution to the GDP, and 74% of cereal output [33]. Despite this magnificent infrastructure, the contribution towards the agriculture employment generation is not considered at all; thus, 31% of the rural people of Punjab live below the poverty line [34]. The total population of Punjab is 110 million, while that of Southern Punjab is about 34.74 million, with an average of five household members [35]. Southern Punjab is one of the country's leading territories in agriculture output, and despite that fact, most of the population lives below the poverty line. Hence, the present study selects Southern Punjab as the population for evaluation of the factors influencing poverty. The recent report published by the Government of Punjab indicates that Dera Ghazi Khan, Faisalabad, Gujranwala, Multan, Sahiwal, and Sargodha are the primary divisions in terms of maize, rice, wheat, cotton, and sugarcane production. The present study considered these divisions and associated tehsils as the population reported in Appendix A Table A6.
The present study used the nonprobability sampling technique in this research to collect data. It is considered to be convenient and suitable for this research. This technique gives accurate results. A close-ended structured questionnaire with a 5-point Likert scale was used for primary data collection (Appendix A, Tables A1-A5). English/Urdu was used as a questionnaire medium, as most farmers cannot read and write the English language. The literature documented the three well-acknowledged methods for the calculation of sample size. The first is the rule of thumb which claims that ten observations are sufficient for each predictor; hence, a total sample of 50 was sufficient as the five predictors in the present study. The G*power calculator was another method used to calculate the minimum sample size in order to validate the findings. The results of the G*power calculator affirmed that a sample size of 166 was sufficient to validate the findings. The third method of sample size calculation was based on the sampling technique. In the case of the probability sampling (simple random sampling) technique, the sample of 384 was sufficient to validate and generalize the findings despite the sample size; hence, the present study considered the sample size of 384 to be sufficient validate and generalizability of findings.

Structural Equation Modelling
Structural Equation Modelling (SME) was employed in order to explain the relationship among multiple variables [36]. It allows one to reveal the structural relationship among the equations (see the conceptual framework in Figure 1). SEM is capable of addressing and assessing the errors in the measurement model. SEM also provides a tool to access the measurement errors which may occur [37]. This approach facilitates multiple regression analysis [38] and multivariable analysis, as it allows one to compute moderating effects and calculate the direct and indirect effects, or even in the case of compounding measurement error when computing interaction terms [39]. PLS-SEM is a type of variance-based structural equation modelling that has been most popular recently [40]. PLS-SEM is a second-generation technique for multivariate data analysis, and is also powered by the features of the first-generation (linear regression, principal components). Furthermore, PLS-SEM is a modelling technique that facilitates the research by measuring the relationship between multiple independent or more dependent variables.
The present study used the PLS-SEM technique to assess the association between latent and measured constructs. The PLS-SEM uses two steps to evaluate the relationship between latent and measured variables. The first step is assessing the measurement model (outer model), and the second is the structural model assessment (inner model). The measurement model assessment, also known as the outer model assessment, specifies the correspondence rules between latent and measured variables [41]. This technique allows the researcher to use as many as several variables for one or multiple numbers of dependent variables. The assessment of the measurement model is based on two key criteria; reliability and validity of items/constructs [42]. The present study assesses the reliability of validity using convergent and discriminant validity. The underpinning objective of reliability assessment is to ensure the stability or consistency of items/constructs. At the same time, validity is used to assess the accuracy of items/constructs for measuring any latent variables [43].

Reliability and Validity
The reliability and validity of constructs/items is a matter of concern which ensured the use of the assessment of the measurement model to proceed with the structural model assessment. Impairment of reliability and validity of constructs/items can mislead the findings and association among the variables. The evaluation of the measurement model is used to assess the reliability and validity of the constructs/items taken under consideration. Reliability refers to the chances/degree of producing similar findings if items are used repetitively to measure the same construct or variables under the same criteria. Validity refers to the degree or chance of measurement of the construct, which intends to be measured by the relevant items [44]. Ensuring the validity of items to measure the required variable is possible if the items measured the same construct, and we could say that the scale is valid. Three tests are frequently used to ensure the validity of the questionnaire in the literature: face/content, construct, and criterion validity.
Face validity is required for the newly developed items; however, if the items are adopted from the previous studies, then there is no need for face validity, as those items were previously used to measure the required constructs [43]. Moreover, content validity ensures that the questions/items used in a questionnaire are adequate to measure the relevant construct. The content validity questionnaire should be duly reviewed by the experts from the academic research. The criterion validity of the questionnaire will be ensured using factor analysis, and factor analysis will provide the relevance of each item being used in the measurement of the relevant construct. The present study used three criteria for assessing the reliability and validity of items/constructs: items loading, composite reliability, and average variance extracted. The threshold values for the factor loadings were 0.50, composite reliability was 0.70, and AVE was 0.50. The study measured sustainable agriculture with four dimensions (environment sustainability, economic viability, production efficiency, and social responsibility) ( Table 1). The loadings of four items for environmental sustainability were less than 0.50, the remaining four items meet the threshold criteria and composite reliability, and AVE meets the minimum threshold of 0.70 and 0.50, respectively. The agriculture production was measured with 15 items, out of which one item failed to meet the threshold value of 0.50, while the remaining items were more than 0.50. As  The agriculture production was measured with 15 items, out of which one item failed to meet the threshold value of 0.50, while the remaining items were more than 0.50. As a result, the values of CR and AVE were more than 0.70 and 0.50. Therefore, the present study measures the government policies with the three items and meets the threshold of 0.50, while the values of CR and AVE were approximately near the threshold values of 0.70 and 0.50 ( Figure 2).  Convergent validity was used to assess the items/construct positively correlations with the same and alternative construct simultaneously. Determination of the convergent validity of the PLS-SEM technique used the average variance extracted (AVE) criteria and outer loadings of items [45]. The average variance extracted was the average variance shared between indicators and latent constructs. In other words, we can say that it was the grand mean value of all the squared loadings of indicators being used to measure a particular construct [45]. The acceptable range of average variance extracted based on the average variance shared by the measured construct must be higher than the variance shared with other constructs in the same model [46]. Discriminant validity is used to assess the uniqueness of a latent construct. Simply put, we can say that a phenomenon captured by an individual construct must be unique and not be captured by the other constructs in the same model [45]. There are various techniques for discriminant validity assessment: cross-loadings among latent constructs, Fornell-Larcker, and Heterotrait-Monotrait criteria. In the initial stage, we assessed that the cross-loading of constructs must be higher at a particular construct than other constructs in the same model [47]. The findings revealed that the diagonal values of the Fornell-Larcker Criterion indicate that all of the values were within the threshold value ( Table 2).
The discriminant validity was assessed by using the loadings and cross-loadings techniques. First, all of the loadings indicate the highest loadings on the relevant constructs. Then, the discriminant validity was rechecked using the Heterotrait-Monotrait Ratio (HTMT). The results reveal that all the values are less than the threshold value of 0.90 (Table 3).

Structural Model Assessment
The independent variables and dependent variables were linked directly. The results were given using the bootstrapping at 5000-sample re-sample criterion and the coefficient beta, standard deviation, t-values, p-values, and f -square [42].

Discussion
The first underlying objective of the present study was to evaluate the association between sustainable agriculture practice and agriculture production. The current study revealed that sustainable agriculture practices are positively and significantly linked with agricultural production. Some of the prior literature has acknowledged that adopting sustainable agriculture practices significantly improves agriculture production [48,49]. The current study's findings aligned well with the existing literature, as developing countries usually cultivate crops using traditional tools and techniques, resulting in lower level of production; thus, adopting sustainable agriculture practices significantly improves agricultural production.
The SDG's goals are to deal with poverty, zero hunger, health, and well-being if the country can produce the required level of agricultural production. The present study's findings aligned well with the existing literature which states that agricultural production positively and significantly influences the reduction in poverty level [50,51]. However, there was a continued argument on topics from the application of foreign assistance to receiving the country's professional assistance. It was debated that the help received complements the restricted local funds for escalation and progress in underdeveloped regions. In contrast, the anti-assistive opine that where external resources flowed, those that were impeded rose, as the resources were majorly transformed into other things, mainly by individual use [52][53][54]. The underlying reason for these findings could be that local farmers considered foreign aid as a tool of restrictions for them. The findings indicate that government policies negatively yet insignificantly influence poverty reduction. This could be because respondents believe that government policies related to the agriculture sector are not supportive, particularly for small-scale farmers, or because the government does not have any policies related to improving agriculture production or reducing poverty. However, the prior literature affirms that government policies are significantly and positively linked with poverty reduction.
By the courage of global partnerships for progress, donors must continue to assist in stimulating local farmers to enhance their agricultural units and farms for optimal productivity. Donors and world organizations must lift the inflow effect of foreign aid in agriculture. Consequently, the donors and government must ensure that foreign assistance is efficiently and effectively consumed for agricultural-related operations in order to escalate regional crop productivity [55]. Such findings could be because the local farmers considered that foreign aid might bring some restrictions along with it.
Government policies negatively yet insignificantly moderate the association between agriculture production and poverty reduction. The findings could be why the respondents believe government policies are against the agriculture sector. Furthermore, agricultural production positively and significantly mediates the association between sustainable agriculture practices and poverty reduction at a 5% significance level. The existing recent literature supports the findings of Sarkar et al. [56] and Sikandar et al. [57].

Conclusions
The contribution of the current study is multifaceted; theoretical, empirical, methodological, and contextual contributions have been made by the current study to the existing body of literature. The underpinning theory claims that a person or a country is poor because they are poor. This means that there is a cyclical relationship between poverty and the poor. The findings significantly contribute to the underpinning theory of the vicious cycle of poverty by considering the underlying variables that can break this circle, including sustainable agriculture practices and agricultural production, towards poverty reduction. Furthermore, the present study considered the role of foreign aid and government policies to be moderating variables. The underlying theory claims that a lack of resources leads to an inability to develop more resources, and this cycle continues indefinitely. The findings contribute to the existing literature and the underpinning theory by explaining the role of sustainable agriculture practices in poverty reduction. Only limited research has been documented so far from the developing countries, which collected the data using survey-based questionnaires from the farmers on the five Likert scales and used the structural equation modelling technique to test the hypothesis.
The study will help the farmers, landlords, policymakers, and regulatory authorities of Pakistan to understand the role of sustainable agriculture practices in improving agricultural production to reduce the poverty level in Pakistan. Furthermore, the findings also help the policymakers and regulatory authorities to understand how farmers perceive foreign aid as well as how they perceive the fact that government policies are not supporting the farmers in Pakistan. This study helps the policymakers and regulatory authorities to understand the significance of government policies, as current farmers consider that current government policies negatively influence the country's agricultural production. The present study's findings help the regulatory authorities, farmers, foreign agencies, and academicians to empirically test the association among the latent constructs. Future studies must also consider other regions or South Asian regions to validate the current study's findings. In addition, future studies need to consider other factors, including the lack of access to finance, market imperfection, storage capacity, infrastructure deficiencies, economic backwardness, and other factors that negatively influence productivity.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Acknowledgments:
The guidance from Hongshu Wang from Northeast Forestry University is greatly appreciated.

Conflicts of Interest:
The authors declare no conflict of interest. Do you feel foreign aid has improved agriculture production in Pakistan?  1  2  3  4  5  10 Do you feel Pakistan relies on foreign aid intervention in the area of agriculture development? 1 2 3 4 5

Appendix A
Note: SD = strongly disagree; D = disagree; SDA = slightly disagree; A = agree; SA = strongly agree. Source: authors' development. 3 Production, processing, and marketing of agricultural products are best carried out at the national and regional level 1 2 3 4 5 4 Technology should be used to make farm labour more rewarding and enjoyable, but not to replace it 1 2 3 4 5 Economic viability 5 Farming is first and foremost a business, like any other business 1 2 3 4 5 6 The primary goal of farmers should be to maximize the productivity, efficiency, and profitability of their farms 1 2 3 4 5 7 The successful farmer is one who earns enough from farming to enjoy a good standard of living 1 2 3 4 5 8 Farmers should purchase most of the goods and services they use on their farm 1 2 3 4 5  9  Large scale farmers can best serve agriculture needs  1  2  3  4  5  10 Farmers should farm only as much land as they can personally care for 1 2 3 4 5

11
The amount of farmland owned by an individual/corporation should be limited in order to encourage land ownership by as many people as possible 1 2 3 4 5 Environmental sustainability 12 Soil and water are the sources of all life and should, therefore, be strictly conserved 1 2 3 4 5 13 Farms should be specialized in one or at most a few crops 1 2 3 4 5 14 The key to agriculture's future success lies in learning to imitate natural ecosystems and farm in harmony with nature Agricultural education programs should teach students about the interrelationships between the environment, agriculture, and people Sustainability is the outcome of the collective decision-making that arises from interaction among stakeholders 1 2 3 4 5 Note: SD = strongly disagree; D = disagree; SDA = slightly disagree; A = agree; SA = strongly agree. Source: authors' development. Crop failure and poor harvest 1 2 3 4 5 5 Farmers incur more costs on agricultural activities 1 2 3 4 5 6 Drying up and rotting of farm produce 1 2 3 4 5 7 Delays planting dates which affect yields 1  2  3  4  5  8 Animal growth, reproduction, and milk production are negatively affected 1 2 3 4 5 9 Pasture, forage, and other animal feeds are negatively affected 1 2 3 4 5 10 Diseases and parasites spread quickly 1 2 3 4 5 11 Reduces animal rate of eating and grazing 1 2 3 4 5 12 Increases animal mortality 1 2 3 4 5 13 Reduction in livestock quality and quantity 1 2 3 4 5 14 Reduction in fish harvest 1 2 3 4 5 15 Displacement of farmers 1 2 3 4 5 Note: SD = strongly disagree; D = disagree; SDA = slightly disagree; A = agree; SA = strongly agree. Source: authors' development. Receiving R&D from the government 1 2 3 4 5 3 Collaboration with government institutions 1 2 3 4 5 Note: SD = strongly disagree; D = disagree; SDA = slightly disagree; A = agree; SA = strongly agree. Source: authors' development. Ability to meet basic needs of shelter, food, and clothing 1 2 3 4 5 3 Ability to access recreational services/facilities 1 2 3 4 5 4 Ability to enjoy luxury goods and services 1 2 3 4 5 5 Increase health and education level 1 2 3 4 5 6 Increase wealth for household members 1 2 3 4 5 7 Increase employment levels and skills 1 2 3 4 5 Note: SD = strongly disagree; D = disagree; SDA = slightly disagree; A = agree; SA = strongly agree. Source: authors' development.