Enhancing the Technical Efficiency (TE) of resources and promoting cost-effective production in agriculture are of paramount importance for various compelling reasons. By harnessing these principles, farmers can significantly improve their agricultural practices, optimize resource utilization, and achieve sustainable and profitable outcomes. One key benefit of improving technical efficiency is increased productivity. By adopting advanced farming techniques, leveraging innovative technologies, and optimizing resource allocation, farmers can maximize their output per unit of input. This not only ensures food security by meeting the growing demand for agricultural products but also contributes to overall economic development. Furthermore, efficient resource management plays a critical role in resource conservation. By practicing technical efficiency, farmers can minimize waste and mitigate environmental impact, thus preserving natural resources for future generations and promoting sustainable agricultural practices [1].

Cost reduction is another pivotal advantage of enhancing technical efficiency and cost-effective production. By optimizing the utilization of resources, farmers can minimize input wastage and reduce production costs [2]. Efficient practices lead to savings in labor, fertilizers, pesticides, and machinery expenses, ultimately improving the profitability and economic viability of agricultural operations. Moreover, cost-effective production enhances the competitiveness of farmers in the market, allowing them to offer competitive prices and withstand market fluctuations. By staying cost-effective, farmers can adapt to changing market conditions and maintain their profitability [3].

Importantly, the benefits of boosting TE and cost-effective production extend beyond the farm gate. These practices have positive spill-over effects on rural development, stimulating local economies, improving living standards, and reducing poverty in rural communities. Enhanced agricultural practices contribute to overall rural development and help create sustainable livelihoods for farmers and their communities, essential for the long-term viability and resilience of agricultural systems in meeting global food demand while minimizing environmental impact. Boosting TE, along with a decline in the Cost of Cultivation (CoC) and Cost of Production (CoP), is essential for increasing productivity, improving profitability, ensuring competitiveness, promoting sustainability, and enhancing farmer resilience in rice farming. These factors collectively contribute to the growth and development of the agricultural sector and the overall well-being of farmers. The Government of Andhra Pradesh initiated the establishment of Rythu Bharosa Kendras (RBKs) on 15th February 2019. Currently, there are 10,778 RBKs operating in the State. These centers perform a wide range of activities, including the sale of pre-tested quality seeds, certified fertilizers, and animal feed. They also offer services such as custom hiring of farm equipment, soil testing with corresponding recommendations for crop selection and fertilizer usage. RBKs facilitate crop insurance payments, provide Minimum Support Prices, procure grains, and facilitate payments to farmers. Through RBKs, farmers gain access to quality inputs, subsidized resources, technical guidance, and knowledge transfer, empowering them to enhance productivity, reduce costs, and improve overall competitiveness in the agricultural sector. These initiatives have a positive impact on technical efficiency, cultivation costs, and production expenses for agricultural commodities, particularly in rice production, which is a staple food crop. Over the past three years, RBKs have played a transformative role in rice farming by promoting sustainable practices, ensuring access to quality inputs, and providing training and advisory services, significantly contributing to enhancing profitability and competitiveness in rice cultivation within the region. Therefore, studying and analyzing the impact of RBKs on TE, CoC, and CoP of rice is important for several reasons. Firstly, understanding the impact of RBKs on TE and cost-effective production helps evaluate the effectiveness of government policies and programs. It allows policymakers to assess the outcomes of RBK interventions and make informed decisions about resource allocation and future agricultural development strategies. Secondly, it helps identify whether RBKs are achieving their intended goals of improving farmers' incomes, reducing production costs, and enhancing their overall well-being. Finally, it allows for a more targeted approach in addressing challenges and providing necessary support to farmers, leading to efficient resource management and increased agricultural productivity. Thus, analyzing the impact of RBKs on TE, CoC, and CoP of rice helps evaluate the effectiveness of agricultural policies, improve resource allocation, enhance farmer welfare, and inform future policy design and implementation in the agricultural sector [4].

Study area

The study was conducted in Andhra Pradesh and Telangana States of India and only in the former State, RBKs are introduced by the Government since 2019 (Figure 1). The sampling procedure involved three stages. Firstly, these two States were purposively selected for the study. Secondly, in Andhra Pradesh, ten RBKs from two predominant rice cultivating districts, namely East Godavari and West Godavari (five RBKs from each district), were purposively chosen [5]. In the third stage, 15 rice cultivating farmers were randomly selected from each RBK, resulting in a total of 150 farmers representing the "treated" category (nt = 150). As for Telangana, two predominant rice cultivating districts, Suryapet and Nalgonda, were purposively selected [6]. From each district, 175 farmers were randomly chosen, making up a total of 350 farmers representing the "untreated" category (nut = 35 0).

Data Envelopment Analysis (DEA)

This technique was employed as a linear programming tool to measure the TE of rice farms in both treated and untreated categories in Andhra Pradesh and Telangana, respectively. The analysis utilized an input-oriented DEA model with Constant Returns to Scale (CRS) based on the works of Banker et al. [7] and Charnes et al. [8]. Let 'N' represent the number of farms or Decision Making Units (DMUs), each of which uses 'K' inputs and produces 'M' outputs. For the 'ith' farm or DMU, these input and output vectors are represented by xi and yi, respectively. The data on inputs and outputs of the 'N' DMUs can be organized into a K × N input matrix, 'X', and an M × N output matrix, 'Y', respectively. To calculate the efficiency score (θ) for the 'ith' DMU, the linear programming problem is solved as follows:

minθλ θ                                                                                                 

Subject to

-yi + Y λ > 0

Θxi - Xλ > 0 (1)

λ > 0

The efficiency score obtained through DEA helps in evaluating the TE of each rice farm, considering the inputs used and the outputs produced. This analysis allows for a comparison between treated and untreated categories in both states, providing valuable insights into the effectiveness of the RBKs and their impact on rice farming practices. The data on the following variables pertaining to Kharif [9] are obtained from a randomly selected rice farms across treated and untreated categories and are subjected to DEA analysis for estimating the TE of rice production. The efficiency analysis has been done using the DEAP version 2.1 program developed by Coelli, [10].

Table 1: Definition of variables used in the DEA model.

Based on Cooper et al. [11], the recommended thumb rule for the acceptable sample size in conducting DEA is expressed as follows: n ≥ max{m×s; 3(m+s)}, where 'n' is the number of units under evaluation, 'm' is the number of input variables, and 's' is the number of output variables. In the present study, with 'm' being 4 (representing the number of input variables) and’s’ being 1 (representing the number of output variables), the sample size used consists of 150 treated rice farms and 350 untreated rice farms. This confirms that the sample size meets the adequacy requirement for conducting DEA, satisfying the thumb rule specified by Cooper et al. [11].Treatment Effects on TE, Coc and Cop of Output.

The impact of participation of rice farmers in RBKs in terms of TE realized, reduction in CoC and CoP of output was studied through estimating different treatment effects models viz., Regression Adjustment (RA), Inverse Probability Weighting (IPW), IPW with Regression Adjustment (IPWRA), and Augmented Inverse Probability Weighting (AIPW) [12]. For the RA estimator, the Average Treatment Effect (ATE) and Average Treatment Effect on the Treated (ATET) were estimated by comparing the differences in the PO means between treatment groups after controlling for covariates through regression analysis. In the IPW model, a two-step approach was also employed. In the first step, the propensity score, which represents the likelihood of receiving the treatment, was estimated using a logistic regression model. In the second step, the inverse of the propensity score was used as a weight in calculating the PO means for each treatment group [13]. The IPWRA (Inverse Probability Weighting with Regression Adjustment) model combines elements of both the IPW and RA methods. The ATE for IPWRA can be specified as follows:

And are attained from the inverse probability-weighted least squares problem for untreated group.

The on the estimated parameters α, β, and X describes the double robustness result:
are the estimated propensity scores. Note that the X’s are a vector of covariates based on observed characteristics.

The AIPW estimator is an IPW estimator that includes an augmentation term that corrects the treatment model when it is mis-specified [14]. This model adds a bias-correction term to the IPW estimator. In this study, ‘S’ represented TE or CoC or CoP as the outcome, ‘Z’ represented farmers’ participation in RBK, denoted the likelihood of farmers participation in RBK considering their individual characteristics, and signified the outcome of the expected value (S) with the assumption that farmers participated in RBK. After the estimation of σ (D) with the logit model, the propensity scores gave an indication of whether farmers participated in RBK or otherwise.

Even if the propensity score was mis-specified, the AIPW still produced a reliable result. Although the propensity score was off, the regression model was still able to forecast the results of farmers who took part in RBK (E[S- p_1 (D)] = 0; E [B-r_1 (M)] = 0). As a result, there was still a prediction of the outcomes of farmers who participated in RBK. As a result, the regression model was written as follows with the numerator adjusted to zero:

The double robustness attribute of AIPW and IPWRA states that only the treatment model or the outcome model needs to be correctly stated for the estimation to be consistent [15]. That is, even if one of the models (treatment or outcome) is mis-specified, the estimators are still consistent. Following variables are considered to estimate the PO means, ATE, ATET, Outcome equations for treated and untreated farms (for TE, CoC & CoP) and Heteroscedastic probit model for RBK participation.

Table 2: Definitions of the variables used in the impact evaluation of RBK.

Policy Analysis Matrix (PAM)

PAM) was employed to assess the domestic and export competitiveness of rice by quantifying the extent of divergence between social and private costs (Table 3). Tradable inputs, including seeds, fertilizers, pesticides, weedicides, and depreciation on machinery, were taken into consideration, while non-tradable inputs encompassed human labor, bullock labor, machine labor, irrigation, farmyard manure, and the imputed rental value of land.

Table 3: Illustrative PAM.

Source: Monke and Pearson [16]

From Table 3, A = private revenue, B = tradable input cost, C = domestic factor cost such as land, labour, capital, etc., D = private profit, E, F, G and H are social values of A, B, C and D respectively. Quantities of inputs and outputs with their respective unit prices were inputted into PAM software, which produced the PAM results. In the table, p_i^p = price of output in private prices, q_i^p = quantity of output in private prices, a_j = tradable input coefficients, p_j^p = price of tradable input in private prices, q_j^p = quantity of tradable input in private prices, b_k = domestic input coefficients, p_k^p = price of domestic input in private prices, q_k^p = quantity of domestic input in private prices, π^p = private profit, p_i^s = output price in social prices, q_i^s = quantity of output in social prices, p_j^s = tradable input price in social prices, q_j^s = quantity of tradable input in social prices, p_k^s = domestic input price in social prices, q_k^s = quantity of domestic input in social prices, π^s = social profit. The items included in the PAM can be used to generate a number of ratios that cast light on domestic competitiveness (w.r.t to Telangana) and export competitiveness (Saudi Arabia and Iran are the leading importers from India) of rice from treated farms compared to untreated counterpart and how these are affected by Government policies [16,17]:

Measures of Private Profitability (Profitability)

Private (Financial) Profits (PP): D = A–(B + C) (7)

Private Cost Ratio (PCR) = C/ (A-B) (8)

Private Cost-Benefit Ratio (PCBR) = (B+C)/A (9)

Measures of Social Profitability (Comparative Advantage or Efficiency)

Social Profits (SP): H = E–(F + G) (10)

Domestic Resource Cost Ratio (DRCR) = G/ (E-F) (11)

Social-Cost Benefit (SCB) = (F+G)/E (12)

Profitability Coefficient (PC) = D/H (13)

Tradable Input Transfers (TIT): J = B–F (14)

Factor Transfers (FT): K = C–G (15)

Measures of Protection Incentives

Nominal Protection Coefficient on Tradable Inputs (NPCI) = B/F (16)

Nominal Protection Coefficient on Tradable Outputs (NPCO) = A/E (17)

Effective Protection Coefficient (EPC) = (A-B)/ (E-F) (18)

Subsidy Ratio to Producers (SRP) = L/E (19)

Measures of International Competitiveness

International Value Added (IVA) (US$) = (E-F)/exchange rate (20)

Coefficient of International Competitiveness (CIC) = G/IVA (21)

The social costs were calculated using the Value Marginal Product approach, which takes into account the factor share (S_i) of various inputs (X_i) along with the mean values of inputs and outputs (Y) and their corresponding prices (P_i). The computation of the social cost of each input is as follows:

Characteristics of Rice Farms across Treated and Untreated Categories:

Table 4 presents a comprehensive overview of the demographic and farm-level characteristics of rice farming households categorized as "treated" and "untreated." The test statistics indicate that there are statistically significant differences in these characteristics between the two categories. Among both treated and untreated farmers, males are the predominant gender engaged in rice farming. The average age of rice farmers, as represented by the FE, is approximately 42 years. Interestingly, untreated farmers tend to have higher farming experience, with an average of 46 years, while the treated category has a lower average farming experience of 33 years. In terms of education, treated farmers have higher levels of formal education, averaging 16 years, compared to the untreated category with an average of six years. The overall average years of formal education for rice farmers in this study is approximately 12 years. Regarding access to agricultural support services, treated farmers have better access compared to the untreated category. Treated farmers have higher percentages of access to extension networks (92%), technical expertise from subject matter specialists (98%), and price information on inputs and outputs (94%). The average LHS for the treated category is 1.08 hectares, which is smaller than the untreated category's average of 1.39 hectares. The overall per capita LHS is relatively small, with an average of 1.22 hectares. Notably, treated farms achieved higher rice yields, averaging 60.8 quintals per hectare, compared to untreated farms, which had an average yield of 52.1 quintals per hectare. This difference in yield may be attributed to treated farmers having relatively better access to quality inputs, TER, and API through support from RBKs. These findings highlight the significant impact of RBK interventions on the characteristics and productivity of rice farming households in the study region.

Table 4: Characteristics of Treated and Untreated Rice Farms.

Note: SD = Standard Deviation, ** & * - Significant at 1 and 5 percent levels respectively.

Overall TE, Pure TE and Scale Efficiency

The results presented in Table 5 reveal that only a small percentage of treated farms (4%) and untreated farms (2%) achieve a perfect TE score of 1.00 (100%) under the assumption of Constant Returns to Scale (CRS). However, a significant proportion of untreated farms (23%) fall into the lowest efficiency group, with TE scores below 60%, which is substantially higher compared to the treated farms (7%). This suggests that the majority of both treated (89%) and untreated (74%) farms have TE scores ranging between 0.61 and less than 1.00, indicating room for improvement in input utilization for rice production. The overall TE scores range between 0.36 and 1.00 for treated farms and between 0.35 and 1.00 for untreated farms, with mean scores of 0.82 and 0.71, respectively. This implies that both treated and untreated farms could potentially reduce their current level of input usage by 18 per cent and 29 per cent, respectively, while maintaining the same level of output. Figure 1 visually depicts the distribution of overall TE scores, showing an asymmetric distribution tilted towards the right side among treated farms compared to the untreated category. This indicates a higher level of overall TE in the treated farms, with a mean score of 0.82. The Pure TE scores, which represent technical inefficiency without considering scale inefficiency, range from 0.29 to 1.00 among the treated category and from 0.35 to 1.00 among the untreated category, with mean scores of 0.89 and 0.82, respectively. Scale efficiency scores, which represent the efficiency in the scale of production, range from 0.39 to 1.00 for treated farms and from 0.38 to 1.00 for untreated farms, with mean efficiency scores of 0.93 and 0.88, respectively [18]. Based on these findings, the following inferences can be drawn:

• The breakdown of the overall TE measure allows for estimating that 11 per cent of pure technical inefficiency and 7 per cent of scale inefficiency exist among treated farms, while the same estimates for untreated farms are 18 per cent of pure technical inefficiency and 12 per cent of scale inefficiency (Figure 2).

Eliminating scale inefficiency could increase the average overall TE from 0.817 to 0.891 for treated farms and from 0.705 to 0.823 for untreated farms. These findings indicate the potential for enhancing technical efficiency and optimizing resource utilization in both treated and untreated rice farms to achieve higher productivity and better overall performance.

Figure 1: Distribution of TE score among treated and control rice farms.

A scale efficiency of 0.932 among treated farms means that majority of them are operating at or near their optimal size compared to the untreated farms, which have a scale efficiency of 0.882. This suggests that treated farms have a better utilization of their production resources in terms of scale efficiency, resulting in more efficient production operations.

The overall technical inefficiency among treated farms is 18 per cent, which is considerably lower compared to the untreated farms with an inefficiency rate of 29 per cent. This indicates that untreated farms are not efficiently utilizing their production resources, leading to sub-optimal output given the available resources. The inefficiency among untreated farms is primarily attributed to improper input use in rice production. Therefore, improving the overall TE scores among untreated farms would require the adoption of practices such as using quality inputs and implementing Good Agricultural Practices (GAPs) to enhance efficiency in input utilization. These findings highlight the need for targeted interventions and support to address inefficiencies in untreated farms, focusing on optimizing input utilization and implementing efficient production practices. By doing so, untreated farms can improve their overall technical efficiency and enhance their productivity, leading to better performance and competitiveness in rice farming. On the other hand, the higher scale efficiency observed among treated farms.

Table 5: Frequency Distribution and Summary Statistics on Overall TE, Pure TE, and Scale Efficiency Measures among Treated and Untreated Categories of Rice Farms.

Table 6: Summary of RTS Results across Treated and Untreated Categories.

Note: Figures in parentheses are percent to total.

Underscores the positive impact of the interventions provided by the RBKs, enabling them to operate more efficiently at their optimal scale.

Figure 2: Percentage of pure and scale in efficiencies in Overall TE among Treated and Untreated rice farms.

Scale of Operations in the Production Frontier

According to Table 6 and Figure 3, approximately 46 per cent of the treated farms and 26 per cent of the untreated farms operate at an optimal scale under CRS conditions. However, a significant proportion of treated farms (23%) and untreated farms (24%) operate at a sub-optimal scale under Increasing Returns to Scale (IRS) conditions. Furthermore, 32 per cent of treated farms and 49 per cent of untreated farms operate at a supra-optimal scale under Decreasing Returns to Scale (DRS) conditions. These findings highlight the potential for improvement in scale size among a considerable portion of both treated and untreated farms. Specifically, 23 per cent of treated farms and 24 per cent of untreated farms have the opportunity to increase their scale size, while 32 per cent of treated farms and 49 per cent of untreated farms have the opportunity to reduce their scale size to enhance resource use efficiencies. However, it is noteworthy that the mean scale efficiencies are already relatively high for both treated farms (0.93) and untreated farms (0.88). This suggests that there is limited scope for further improvement in farm size to increase efficiencies among both categories of farms [18]. In general, inefficiency in resource usage can be attributed to either inappropriate scale or misallocation of resources. In this study, since the mean scale efficiencies are considerably high for both treated and untreated farms, it indicates that the inefficiency in resource utilization is primarily due to improper input use rather than scale inefficiency. Additionally, the mean pure technical efficiency scores are lower than the scale efficiency scores for both categories of farms. This further supports the notion that the inefficiency in resource utilization is primarily a result of managerial inefficiency rather than scale inefficiency. These findings suggest that addressing managerial inefficiency and improving the allocation of inputs would be crucial for enhancing resource use efficiencies among both treated and untreated farms. By focusing on optimizing input utilization and implementing efficient farming practices, farmers can improve their overall technical efficiency and achieve higher productivity in rice production.

Figure 3: Distribution of scale operating among treated and untreated rice farms.

Double Robust Estimation of Treatment Effects

To address the endogeneity problem, the study employed a combination of different treatment effect models, namely, RA, IPW, IPWRA, and AIPW. In order to obtain more accurate standard errors when using these models, the study utilized Bootstrap methods with 1000 replications. This approach helps in accounting for the uncertainty and variability in the estimation, providing more robust and reliable results. The density plots shown in Figure 4 illustrate the estimated probabilities of treatment for both the treated and untreated farms. Notably, none of the treatment units have estimated probabilities at the extreme points of '0' or '1', indicating that there is no perfect separation between the treated and untreated groups. Furthermore, there is considerable overlap of the density curves, indicating that the treatment and untreated groups share similar characteristics. This observation is crucial as it supports the validity of the doubly robust models (IPWRA and AIPW) and indicates that the estimation is less prone to biases caused by sample self-selection and unobservable factors. Thus, the combination of treatment effect models and the utilization of Bootstrap methods help mitigate potential biases arising from endogeneity issues, allowing for more reliable inference in the study.

Figure 4: Density of overlap of propensity score among treated and untreated rice farms.

Table 8 presents the estimated ATE and ATET resulting from farmers' participation in RBKs, as estimated by different treatment effect models. The ATE estimates, which represent the differences in means between treated and untreated farms, consistently show similar signs, magnitudes, and levels of significance across the different models. This indicates that farmers' participation in RBKs has a substantial positive influence on resource use efficiency (TE), leading to a decline in the CoC and CoP among the treated farms compared to the untreated farms. The ATE values from the RA, IPW, IPWRA, and AIPW models all suggest that rice farmers

This study presents a comprehensive analysis of the impact of farmers' participation in RBKs on resource use efficiency (TE) and the cost of cultivation in rice farming. The consistent findings indicate that participating in RBKs positively influences TE and cost parameters compared to untreated farms. The results highlight that both treated and untreated farms have room for improvement in input utilization for rice production. While only a small percentage of farms achieve maximum TE scores of 1.00 under the assumption of CRS, a significant proportion of untreated farms (23%) fall into the lowest efficiency group with TE scores below 60 percent. This suggests considerable potential for improving input utilization for the majority of farms in both categories. The overall TE scores indicate that both treated and untreated farms could potentially reduce their current level of input usage by 18 percent and 29 percent, respectively, while maintaining the same level of output. Notably, the presence of RBKs has minimized input slacks among treated farms, indicating the effectiveness of RBKs in reducing input wastage and improving resource use efficiency. To ensure the reliability of the findings and address potential issues of sample self-selection and unobservable factors, the study employs various treatment effect models such as RA, IPW, IPWRA, and AIPW. The results consistently demonstrate that farmers' participation in RBKs has a substantial positive impact on TE scores, CoC, and CoP. These impacts are economically significant and contribute to the domestic and export competitiveness of the rice industry in the region. In terms of domestic competitiveness, the study compares the rice industry in Andhra Pradesh with that of Telangana, considering factors such as production costs, input accessibility and pricing, infrastructure, government policies, and market conditions. This information enables policymakers and stakeholders to assess the relative competitiveness of the rice industry between the two regions. Similarly, the study examines the export competitiveness of rice from Andhra Pradesh to leading importing countries. This analysis involves evaluating factors such as production costs, quality standards, transportation and logistics, trade policies, market access, and global market conditions. The findings show that treated farms exhibit greater export competitiveness compared to untreated farms and are also better protected, as they pay less for domestic factors obtained from RBKs, resulting in higher social profits. Overall, the study highlights the positive impact of RBKs on TE and cost-effective production in the rice farming sector. The findings provide policymakers and stakeholders with valuable insights to understand and improve the relative competitiveness of the rice industry, formulate appropriate strategies, and foster sustainable growth in the sector. By leveraging the benefits of RBKs, Andhra Pradesh can further enhance its domestic and export competitiveness, leading to a stronger rice industry and increased economic benefits for the region.

1. Monika G, Sanjay KJ and Chandra K. Role of RBKs in Supply chain of sorghum in Andhra Pradesh the Pharma Innovation Journal. 2022; 11: 2461-2463.
2. Aritri C. Impact of Information on Technical Efficiency of Agricultural Production in India BASEU Working paper series. 2022; 14.
3. Heera SN, Kumar KNR. Economic analysis of groundnut farmers participating in Rythu Bharosa Kendras (RBKs) in Guntur district of Andhra Pradesh. The Pharma Innovation Journal. 2023; 12: 5586-5590.
4. Anantha DRK, Bala HRP, Sailaja V and Shaik NU. Awareness of Farmers towards Rythu Bharosa Kendras (RBKs) Products and Services in Prakasam district of Andhra Pradesh. The Journal of Research ANGRAU. 2022; 50: 144-148.
5. Statistical Abstract. Government of Andhra Pradesh 2021.
6. Telangana State Statistical Abstract. Government of Telangana 2022.
7. Banker RD, Charnes A, Cooper WW. Some Models for Estimating Technical and Scale Inefficiencies in Data Envelopment Analysis. Management Science. 1984; 30: 1078–1092.
8. Charnes A, Cooper WW, Rhodes E. Measuring the efficiency of decision making units. European. Journal of Operations Research. 1978; 2: 429–444.
9. Coelli TJ. A Guide to DEAP Version 2.1: A Data Envelopment Analysis (Computer Program). Center for Efficiency and Productivity Analysis Department of Econometrics. University of New England. Armidale NSW 2351 Australia.
10. Cooper WW, Seiford LM, Tone K. Data envelopment analysis: a comprehensive 2007.
11. Wooldridge JM. Econometric Analysis of Cross Section and Panel Data. 2nd edition. Cambridge MA: MIT Press.2010.
12. Imbens GW. Nonparametric estimation of average treatment effects under exogeneity: A review. The Review of Economics and Statistics. 2004; 86:4–29. 
13. Rubin DB, van der Laan MJ. Empirical Efficiency Maximization: Improved Locally Efficient Covariate Adjustment in Randomized Experiments and Survival Analysis. Int J Biostat. 2008; 4.
14. Andrea NTP. Market-based school reforms and educational inequality: the case of Chile Unpublished PhD thesis submitted to Nuffield College University of Oxford Michaelmas Term. 2020.
15. Monke E, Pearson SR. The Policy Analysis Matrix for Agricultural Development. Cornell University Press Ithaca. 1989.
16. Masters WA. A Winter-Nelson. Measuring the Comparative Advantage of Agricultural Activities: Domestic Resource Costs and the Social Cost-Benefit Ratio. Am J Agric Econ. 1995; 77: 243-50.
17. Tipi T, Yildiz N, Nargeleçekenler M, Çetin B. Measuring the technical efficiency and determinants of efficiency of rice (Oryza sativa) farms in Marmara region Turkey. New Zealand. Journal of Crop and Horticultural Science. 2009; 37: 121–129.
18. Khondoker AM, Roderick MR, Murty MVR, Samarendu M and Tao Li. Benefits of the development and dissemination of climate-smart rice: ex ante impact assessment of drought-tolerant rice in South Asia Mitigation and Adaptation Strategies for Global Change. 2017; 22: 879-901.
19. Adeyemo T, Okoruwa V and Akinyosoye V. Estimating causal effects of cassava based value-webs on smallholder’s welfare: a multivalued treatment approach 30^th International Conference of Agricultural Economists. 2018 July 28^th – August 2018. Vancouver.
20. Dar MH, Waza SA, Shukla S, Zaidi NW, Nayak S, Hossain M, Kumar A, Ismail AM, SinghUS. Drought tolerant rice for ensuring food security in Eastern India Sustainability. 2020; 12: 2214.
21. Fasakin IJ, Ogunniyi AI, Bello LO, Mignouna D, Adeoti R, Bamba Z, Abdoulaye T and Awotide BA. Impact of Intensive Youth Participation in Agriculture on Rural Households Revenue: Evidence from Rice Farming Households in Nigeria 2021; 12: 584.
22. Richard KB, Kyeremeh OH, Bismark A, Lesley H and Samuel KCK. The Nexus between Cocoa Farmers’ Business Schools Participation and Impact to Support Livelihood Improvement Strategies in Ghana. 2021; 1-16.
23. Sseguya H, Robinson DS, Mwango HR, Flock JA, Manda J, Abed R. The impact of demonstration plots on improved agricultural input purchase in Tanzania: Implications for policy and practice. 2021; 16.
24. Baruwa OI, Masuku MB and Alimi T. Economic analysis of plantain production in derived savannah zone of osun state Nigeria. Asian Journal of Agricultural Sciences. 2011; 3: 401-407.
25. Kainga PE, Seiyabo IT. Economics of plantain production in Yenagoa local government area of Bayelsa State.  Journal of Agriculture and Social Research. 2012; 12: 114-123.  
26. Liverpool LSO, Gbolagade BA, Oyeleke RO. Enhancing the competitiveness of agricultural commodity chains in Nigeria: Identifying opportunities with cassava rice and maize using a Policy Analysis Matrix (PAM) framework. International Food Policy Research Institute (IFPRI). 2009.
27. Ugochukwu AI and Ezedinma Intensification of rice production systems in South-Eastern Nigeria: a policy analysis matrix approach. International Journal of Agricultural Management and Development. 2011; 1: 89–100.
28. Kanaka S and Chinnadurai M. The policy analysis matrix of rice cultivation in India. European Journal of Basic Applied Science. 2015; 2: 33–45.
29. Ahmed NSE and Elrasheed MMM. Profitability and Competitiveness of the Main Crops Grown under Rain-Fed Sector of Gadarif State Sudan. Asian Journal of Agricultural Extension. Economics and Sociology. 2016; 11: 1–7.
30. Saptana ALS,  Perwita AD, Bambang S,  Endro G, Syahrul GS, Hayati NQ, Yusuf S,  Muhammad PY, Sudi M and Dian PA. Analysis of competitive and comparative advantages of potato production in Indonesia. 2022.
31. Oluwafunmilola FA, Adewumi TA, Adebayo SB, Ousmane C and Asiedu Competitiveness of the Food Crop Production System in Nigeria: A Policy Analysis Matrix Approach. Journal of Competitiveness Studies. 26: 162-182. 
32. Soetriono DS, Ariq DM, Dimas BZ. Proceeding from EDP Sciences: The competitiveness of Pronojiwo snake fruit. 2020; 142: 05007.
33. Avane A, Amfo B, Aidoo R and Mensah OJ. Adoption of organic fertiliser for cocoa production in Ghana: Perceptions and determinants. African Journal of Science Technology. Innovation and Development. 2022; 14: 718–729.
34. Cooper TJ. A Guide to DEAP Version 2.1: A Data Envelopment Analysis (Computer Program). Center for Efficiency and Productivity Analysis Department of Econometrics. University of New England Armidale. 2351 Australia. 1996.
35. Glynn A, Quinn K. An introduction to the augmented inverse propensity weighted estimator. Political Analysis. 2010; 18: 36–56.
36. Gulati A. and Kelley T. Trade Liberalization & Indian Agriculture. Oxford University Press New Delhi India. 2000
37. Imbens GW. The role of the propensity score in estimating dose-response functions. Biometrika. 2000; 87: 706–710.
38. Imbens GW. And Jeffrey MW. Recent developments in the econometrics of program evaluation. Journal of Economic Literature. 2009; 47: 1: 5-86.
39. Ma W, Zhu Z, Zhou X. Agricultural mechanization and cropland abandonment in rural China. Applied Economics Letter. 2021; 29.
40. Nelson CG and Martin P. The Costs of Indonesian Sugar Policy: A Policy Analysis Matrix Approach American Journal of Agricultural Economics. 1991; 704.
41. Simar LL and Wilson PW. Two-stage DEA: Caveat emptor. Journal of Productivity Analysis. 2011; 36: 205–218.
42. Rahman S. Sujan MD. Debasish CA.
43. Kabir RR and Rahman M. Intensity of adoption and welfare impacts of drought-tolerant rice varieties cultivation in Bangladesh. 2022; 8: 1-8.