Impact of combined climate-smart agriculture practices on the technical efficiency and efficiency yield gap in Ethiopia Authors: Yadeta Bedasa, Adeba Gemechu, Amsalu Bedemo Published: 2025 ABSTRACT Smallholder farmers in Ethiopia rely on family labour and lack the efficient skills necessary to produce crops. They frequently own less than one hectare of land and use animal draft power to cultivate the soil. Rainfall serves as the primary source of water for crops in agriculture. This study aims to investigate the impacts of crop diversification, agroforestry, and adjusted planting dates on the efficiency of farmers. A multinomial endogenous switching regression (MESR) model was used to analyse these factors. Multistage sampling is used to obtain cross-sectional data from 385 randomly selected households. The results revealed that adopting adjusted planting dates, crop diversification, and agroforestry significantly increased technical efficiency by 25%, 57%, and 54%, respectively. Adopting an adjusted planting date and agroforestry decreases the efficiency yield gap by 2.52 t/ha and 1.16 t/ha, respectively. Compared to nonadoption, adopting crop diversification, agroforestry, and adjusting planting dates results in a greater average technical efficiency score and a smaller average efficiency yield gap per hectare. CSA practices operate best in combination with high farmer efficiency, which is crucial for realizing their highest productivity potential. However, in scenarios where farmers do not possess the technical capability to be effective, the adoption of climate-smart agriculture (CSA) can be compromised. On this basis, we recommend to policymakers that they give top priority to measures that enhance farmers’ technical efficiency as a primary cornerstone of CSA policy. Introduction The main topic of this suggested study is Ethiopia’s large productivity gap. Ethiopia has failed to meet its genetic potential in terms of maize yield productivity. Ethiopia continues to import maize and produces far less than it might, which is the primary driving force behind this study. Poor soil fertility, which is caused mostly by ongoing cropping without restoring depleted minerals, is the primary issue raised as the foundation for this investigation1. According to studies by2,3, the average annual rainfall has fallen in western Oromia, whereas temperatures have increased. The maximum temperature, as well as the seasonal and yearly minimum temperature trends, all pointed to significant warming in the study area3. The studied climate data demonstrate variations in the length of the growing season, the start date, and the end date, as well as a change in climate. On the basis of historical climate data, the growth season began in early June, and the Kiremt rains ended in the middle of October2. In the last 30 years, the observed average total annual rainfall over western Oromia has shown a significant downwards trend (43.67 mm/decade), with relatively high interannual variability, whereas the maximum, mean, and minimum temperatures have increased at rates of 0.22 °C and 0.218 °C/decade, respectively4,5. According to the4 study, the yearly rainfall has been decreasing, whereas the mean maximum and lowest temperatures have shown a statistically significant upwards trend. On a seasonal basis, the main rainy season (June–August) is characterized by a decreasing tendency in rainfall, whereas the autumn season (September–November) has an increasing trend. According to studies by5,6, the unpredictable rainfall patterns, increases in temperature, decreases in rainfall, and shifts in the duration of rain in the study area are all signs of climate change. The results of the study by5 revealed that during the past three decades, crop yields were impacted by the increased frequency and intensity of drought events caused by rising temperatures and falling precipitation. The crop output of western Oromia has decreased due to climate change. Climate adaptation and mitigation strategies have enormous potential to reduce its long-term vulnerability and food insecurity. In the previous three decades, there were four occurrences of moderate to severe agricultural drought intensity, according to a study by5. There was a strong relationship between crop yield and drought indicators. However, Western Ethiopian agriculture is subsistence-based and depends on rainfall. Small-scale farmers who employ traditional farming methods and use traditional farming equipment produce the majority of products. The study area is predicted to experience some negative effects of climate change, particularly in the agricultural sector, which is a major source of income for many people living there. According to research by3, climate change has a negative impact on rural areas’ crop yields (49%), increases pest and disease rates (34%), and increases soil erosion rates (98%). This may worsen current food insecurity. Furthermore, a recent study by3,7,8, and9, which was performed in western Oromia, revealed that due to climate change-related factors, agricultural productivity occasionally decreased. Changes in the beginning of the farming season have shown the effects of climate change on crop production6. Climate change is thought to have the greatest impact on crop yields. Analysis of historical climate data reveals a shortened growing season, attributed to the late arrival and premature cessation of rainfall. The western part of Ethiopia is already facing considerable challenges from the current climate variability, including natural disasters, resource degradation, impacts on food security, water and energy availability, poverty alleviation, and sustainable development initiatives. Despite these problems, climate smart agriculture (CSA) practices have shown promise for the future10. CSA is a strategy that sustainably changes and shifts agricultural production by increasing productivity, enhancing resilience, and reducing greenhouse gas emissions to achieve food security in light of climate change11. Climate-smart practices in farming contribute to increased adaptation capacity by making efficient use of resources and creating agricultural systems that are able to resist the effects of climate change12. Moreover, numerous studies suggest that the best chances of increasing agricultural productivity to feed the expanding population depend on the application of CSA13,14. A study by15 revealed that agricultural productivity increases by 29–126% when households adopt at least one CSA strategy. Thus, CSA continues to improve and increase the yield of farms. Moreover, CSA strategies can potentially minimize soil loss in agricultural fields by up to 4% compared with a climate change scenario without implementing these strategies16. However, the acceptance and continuous usage of CSA technologies have fallen below expectations. The adoption of CSA practices also remains low in Ethiopia17,18. In addition, the inefficiencies of smallholder farmers in the production of maize exist due to problems such as pest outbreaks, weather uncertainty, and a lack of knowledge and expertise in agronomic methods and land preparation19. However, a study by8,20 revealed that in western Oromia, deforestation, a lack of crop insurance, a lack of training, a lack of weather forecasting and early warning systems, a lack of funding, a high cost of farm inputs, and reliance on a single rainy season were the main obstacles preventing farmers from adapting to climate change. A previous study confirmed that adopting the CSA practice leads to an increase in the technical efficiency score of maize producers. The technical efficiency of farmers increases with the implementation of CSA21,22. According to23, technical efficiency significantly increased with the adoption of CSA practices, either separately or in combination. Furthermore, the adoption of improved maize varieties and soil and water conservation measures resulted in 15% and 9% increases in farmers’ technical efficiency, respectively24. The smallholder farmers in the study area are inefficient in terms of crop production, but little research has been conducted on how CSA impacts farmers’ technical efficiency scores in western Ethiopia. Furthermore, prior studies on the effect of CSA on the efficiency yield gap in Ethiopia are inadequate. Several scholars have examined CSA adoption and its effects on maize yield. For example25, conducted a study in Ethiopia’s northwestern highlands, a study by26 examined the Teleyayen Basin in northeast Ethiopia, and27 conducted research in the southern Rift Valley of Ethiopia. Western Ethiopia, the area of interest in this study, has received relatively little scholarly attention. The study by28 examined land use and cover dynamics and their implications in the Alwero Watershed, western Ethiopia. The study by29 employed a modelling and forecasting approach to analyse rainfall and temperature trends in the East Wollega Zone, western Ethiopia. A study by30 evaluated average yearly soil loss rates and prioritized them at the subwatershed level. A study by31 investigated spatial‒temporal variability and trends in rainfall and temperature in the Alwero watershed, western Ethiopia. Previous research has investigated the phenomenon of climate change, including its impact on farmers, their awareness of the issue, and strategies for adaptation and mitigation26,31–35. Research on the economic and environmental advantages of CSA has utilized the WOCAT database, which includes studies28,36,37, and38. The research presented in39 examines the relationship between crop yield and net value. CSA contributes to reducing the yield gap40–44. It is unknown how much the efficiency yield gap in t/ha would be reduced if average farmers in the research area adopted and implemented CSA in each practice. To implement effective policy interventions for maize yield improvement, comprehending the contribution of CSA to bridging the efficiency yield gap is essential. By thoroughly examining the effects of CSA on efficiency yield gaps and offering potential policy solutions, the current study is expected to fill a research gap. The first component of the study examines the magnitudes of the efficiency yield gaps in maize-based agricultural systems in western Oromia. The second component of the study examines the impact that CSA practices have on the efficiency yield gap and technical efficiency score of maize producers. The study’s results provide crucial information that will enable the government and other stakeholders to make well-informed decisions. A practical strategy for increasing agricultural productivity and closing the yield gap is needed to address the issues of food insecurity and poverty in the study area. This study employed quantitative methodologies to illustrate the phenomenon under investigation. Stochastic frontier analysis is used to estimate and explain the efficiency yield gap for maize farming in western Oromia. Cross-sectional data for the years 2023–2024 were employed to achieve the research objectives. Methods Description of the study area The research was conducted in the western Oromia region. Figure 1 displays a map of the research geographical area. The data used in this study were collected from 385 maize producers in three major potential maize-growing districts (Sibu Sire, Gudeya Bila, and Gobu Seyo) during the 2023/24 cropping season. Agriculture serves as the main source of income and means of livelihood in the studied area. The farming method is characterized by the production of livestock and mixed crops at the subsistence level. Most crop cultivation is rain-fed, with the exception of a few locations that use traditional, small-scale irrigation. Fig. 1Map of the study area.Source: ETHIO-GIS Map (2023). Map of the study area. Source: ETHIO-GIS Map (2023). Sampling design The study utilized data gathered through a multistage sampling process. The process of selecting (1) sample districts, (2) kebeles, and (3) smallholder farmers comprises the following three stages. Stage 1: In the first stage, three districts, namely, Sibu Sire, Gudeya Bila, and Gobu Seyo, were selected from the maize-growing districts of western Oromia. Stage 2: In the second stage, maize-growing Kebeles were selected from districts proportionally. Stage 3: In the third stage, 385 household heads were randomly selected from the total selected kebeles in proportion to the population. The formula from45 was used to calculate the sample size of the respondent farm households. The Cochran formula is a statistical tool used to determine the appropriate sample size for a study, factoring in the desired level of confidence, the desired level of accuracy, and the estimated proportion of the attribute within the population. For a large population, the Cochran formula is appropriate. The sample size for the large population proportion was determined via an equation devised by46.2.1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{n}=\frac{{\mathrm{n}}_{0}}{1+\frac{{(\mathrm{n}}_{0}-1)}{\mathrm{N}}}$$\end{document}2.2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{n}}_{0}=\frac{{\mathrm{Z}}^{2}\mathrm{p}\mathrm{q}}{{\mathrm{d}}^{2}}$$\end{document} N = Total number of the population. no = the desired sample size when the population is greater than 10,000. n = Sample size when the population is less than 10,000. Z = 95% confidence limit, i.e., 1.96. p = 0.5 (proportion of the population to be included in the sample, i.e., 50%). q = 1–0.5, i.e., (0.5). d = margin of error or degree of accuracy desired (0.05). The sample size was calculated as follows: no = ((1.96)2 *(0.5) *(0.5))/(0.05)2=385. The following formula is used to determine the proportional sample size from each district:2.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{n}\mathrm{i}=\frac{\mathrm{n}\mathrm{*}\mathrm{N}\mathrm{i}}{\mathrm{N}}$$\end{document} where ni = the sample size for each kebel, Ni is the overall household head count for each kebel, N is the overall household head count for the three districts that were chosen, and n is the overall sample size for the districts that were chosen. A total of 385 respondents were chosen by the researcher via proportionate sampling. An appropriate number of respondents from the selected kebeles was obtained via a simple random sample technique. This method of sampling was chosen because it ensures that every member of the population has a fair or equal chance of being chosen. Data Sources, types and data collection methods In this study, both qualitative and quantitative data were collected from primary and secondary data sources. The community-level qualitative data were collected through focus group discussions, key informant interviews, and observations. The primary data were collected via a structured questionnaire that was administered by trained enumerators before the actual data were collected. Documents from the district’s office of agriculture and other pertinent entities provided secondary data. Ethical approval or consent Written informed consent was obtained from all participants prior to their involvement in the study. Permission to conduct the interviews for the purposes of this research was obtained from all the respondents, who were fully informed about the purposes of this research and how their responses would be used and stored. Survey or Observational Study - Informed Consent: All procedures performed in this study involving human participants were conducted ethically according to the ethical standards of the Institutional and/or National Research Committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study. Study involving Identifiable Human Data or Samples - Consent: Research involving the use of human data/samples was approved by the Institutional Review Board of Wollega University (Approval No. 208). Informed consent for the use of their data/samples was obtained from all participants. Definitions and concepts of adjusting plant date, agroforestry, and crop diversification CSA techniques exist worldwide, but they are selected on the basis of the consequences of climate change. Western Ethiopia has been impacted by climate change since rainfall seasonality has changed over the last 20 years, resulting in changes in the duration, start date, and end date of the agricultural season. Therefore, agroforestry, crop diversification, and adjusting planting dates to align with shifting rainfall patterns are some of the strategies farmers in western Ethiopia have employed in response to climate change. Adjusting the plant date The adjustment of plant date refers to the deliberate alteration of crop planting or sowing dates to align with changing rainfall, temperature, and seasonal patterns11,47., and48 reported that altering crop planting dates was the most popular adaptation choice among sample farm households in northern Ethiopia (89%), southwest Ethiopia (47%), and central Ethiopia (45.59%), respectively. Planting 4–6 days earlier in the growing season resulted in an average yield increase of 4.9–66.9%11,49. reported that planting maize cultivars between 15 May and 1 June resulted in relatively high yields under high-altitude agroecological conditions50. reported that delaying sowing by 1–2 weeks can successfully minimize the negative impacts of climate change. Agroforestry Agroforestry involves incorporating trees into a land-use management system alongside crops and animals on the same plot of land. This situation creates sustainable farming systems that improve productivity, increase soil fertility, preserve biodiversity, and provide resilience to environmental hazards such as soil erosion and climate change by combining forestry and agricultural practices. Home gardens, nature preserves, border planting, forest plots, and coffee-based and fruit-based systems are examples of traditional agroforestry systems in Ethiopia51. Agroforestry has positive implications, including reduced fertilizer and pesticide runoff; increased soil organic matter, soil fertility, and carbon storage; water regulation and quality; and insect and disease control51–5353. reported that agroforestry systems reduce soil, SOC, N, P, and K losses by 27–76%, 21–78%, 20–82%, 24–82%, and 22–84%, respectively. Compared with the tree/hedgerow-free maize monoculture, agroforestry increased the average maize yield by 0.24 Mg ha-1 (7%)54. Agroforestry can improve soil fertility and recover biomass carbon in the range of 12–228 Mg ha⁻151. Crop diversification Crop diversification is used to describe a farming practice where farmers plant many crop varieties on a single plot of land in a single growing season. Crop diversification has a significant effect on agricultural yield and productivity. Some explanations for this influence include enhanced resilience to climate fluctuations, better pest and disease control, and improved soil health. Crop diversification has several advantages, including the ability to better manage pests and diseases as a crop failure mitigation strategy, control weed populations, improve soil health, increase resilience to abiotic stresses such as droughts or extreme temperatures, and disrupt the life cycles of pests that thrive on specific plants55–57. Cereal cultivation with species diversification exceeded maize‒soybean rotation in terms of productivity, with an average 37% greater yield58. Compared with maize monocultures, maize intercropped with the Gora faba bean variety at a 50% density increased overall maize yields by 13%59. Methods of data analyses Descriptive statistics In this study, descriptive statistics, including frequency, percentage, graphs, figures, and tables, were used to summarize and present demographic and socioeconomic data. Econometric analysis Estimation of outcome variables The technical efficiency (TE) and efficiency yield gap, the study’s outcome variables, are estimated via a stochastic frontier model. The dependent variable of the estimated production function was the maize yield in t/ha. The translog production function with fourteen input variables is as follows: fertilizer, oxen, labor, seed, fertilizer × fertilizer, fertilizer × oxen, fertilizer × labor, fertilizer × seed, oxen × oxen, oxen × labor, oxen × seed, labor × labor, labor × seed, and seed × seed. The model’s input and output variables were derived from the research presented in60–64. However, maize output in kg was employed as the dependent variable, and land was one of the explanatory variables in the studies by65–67, and68. The translog model is a widely used statistical tool for modelling crop responses to fertilization, demonstrating superior predictive accuracy compared with other functional methods. The fertilizer utilization rate is thought to influence the maize yield response, and this relationship can be approximated with a second-order polynomial function. This model, the most widely employed second-order flexible functional form, has a significant number of applications69,70. Technical efficiency (TE) is the ratio of the observed yield (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{Y}}_{\mathrm{i}}$$\end{document}) to the corresponding frontier yield (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\mathrm{Y}}_{\mathrm{i}}}^{\mathrm{*}}$$\end{document}), which is conditional on the levels of inputs used by smallholder farmers. The efficiency yield gap is the difference between the technically efficient yield (YTEx) and actual yield (Ya). Technically efficient yields (YTEx) refer to the highest possible yields obtained given the observed levels of inputs in a well-defined biophysical environment. YTEx can be quantified with methods of frontier analysis. The efficiency yield gap measures the extent to which farmers can produce more by using the same inputs under the same production conditions. The translog production function with fourteen input variables is specified as follows:2.4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{l}\mathrm{n}{\mathrm{Y}}_{\mathrm{i}}={\beta\:}_{0}+\sum\:\left({\beta\:}_{i}*ln{x}_{i}\right)+\left(\frac{1}{2}\right)\mathrm{*}\sum\:\sum\:\left({\beta\:}_{ij}*ln{x}_{i}*ln{x}_{j}\right)+{\mathrm{v}}_{\mathrm{i}}-{\mathrm{u}}_{\mathrm{i}},\:\mathrm{i}=\mathrm{1,2},3,\dots\:\mathrm{n}\:\mathrm{a}\mathrm{n}\mathrm{d}\:{\mathrm{u}}_{\mathrm{i}}\ge\:0$$\end{document} where: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:\mathrm{l}\mathrm{n}{\mathrm{Y}}_{\mathrm{i}}$$\end{document} is the log of maize yield. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:ln{x}_{i}$$\end{document} is represents the log of inputs. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\beta\:}_{0}$$\end{document} is a constant term. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\beta\:}_{i}$$\end{document} is the coefficient for the ith input. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\beta\:}_{ij}$$\end{document} is the coefficient for the interaction between input i and j. The double summation ΣΣ indicate summation over all possible pairs of inputs. The term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left(\frac{1}{2}\right)\mathrm{*}\sum\:\sum\:\left({\beta\:}_{ij}*ln{x}_{i}*ln{x}_{j}\right)$$\end{document} captures interaction effects between inputs. where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:\mathrm{l}\mathrm{n}{\mathrm{Y}}_{\mathrm{i}}$$\end{document}is the natural logarithm of the maize yield; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{x}}_{\mathrm{i}}$$\end{document} represents the input variables; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\upbeta\:}$$\end{document} represents the slope of the input variables; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{v}}_{\mathrm{i}}$$\end{document} represents a random error term; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{u}}_{\mathrm{i}}$$\end{document} repres000000ents the inefficiency term; and n represents the number of observations. Technical efficiency (TE) is the ratio of the observed yield (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{Y}}_{\mathrm{i}}$$\end{document}) to the corresponding frontier yield (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\mathrm{Y}}_{\mathrm{i}}}^{\mathrm{*}}$$\end{document}), which is conditional on the levels of inputs used by smallholder farmers. Thus, the technical efficiency of each smallholder farmer can be given as:2.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{T}\mathrm{E}}_{\mathrm{i}}=\frac{{\mathrm{Y}}_{\mathrm{i}}}{\mathrm{f}\left({\mathrm{X}}_{\mathrm{i}},\:{\upbeta\:}\right)\mathrm{e}\mathrm{x}\mathrm{p}\left\{{\mathrm{v}}_{\mathrm{i}}\right\}}=\frac{\mathrm{f}\left({\mathrm{X}}_{\mathrm{i}},\:{\upbeta\:}\right)\mathrm{e}\mathrm{x}\mathrm{p}\left\{{\mathrm{v}}_{\mathrm{i}}-{\mathrm{u}}_{\mathrm{i}}\right\}}{\mathrm{f}\left({\mathrm{X}}_{\mathrm{i}},\:{\upbeta\:}\right)\mathrm{e}\mathrm{x}\mathrm{p}\left\{{\mathrm{v}}_{\mathrm{i}}\right\}}=\mathrm{e}\mathrm{x}\mathrm{p}\left\{-{\mathrm{u}}_{\mathrm{i}}\right\}$$\end{document} The technical inefficiency is writtenas2.6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{1-\mathrm{T}\mathrm{E}}_{\mathrm{i}}$$\end{document} The model can be estimated via the maximum likelihood technique, and one can find consistent estimators of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\upsigma\:}}^{2}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\upgamma\:}$$\end{document} such that2.7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\upsigma\:}}^{2}={{{\upsigma\:}}_{\mathrm{u}}}^{2}+{{{\upsigma\:}}_{\mathrm{v}}}^{2}$$\end{document}2.8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\upgamma\:}=\frac{{{{\upsigma\:}}_{\mathrm{u}}}^{2}}{{{\upsigma\:}}^{2}}=\left.\frac{{{{\upsigma\:}}_{\mathrm{u}}}^{2}}{{{{\upsigma\:}}_{\mathrm{u}}}^{2}+{{{\upsigma\:}}_{\mathrm{v}}}^{2}}\right\},\:0<{\upgamma\:}<1$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{{\upsigma\:}}_{\mathrm{u}}}^{2},{{{\upsigma\:}}_{\mathrm{v}}}^{2}\:\mathrm{a}\mathrm{n}\mathrm{d}\:{{\upsigma\:}}^{2}\:\:$$\end{document} represent a deviation from the frontier due to inefficiency, noise, and total deviation from the frontier, respectively. The total variation in yield from the frontier yield is given by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\upgamma\:}$$\end{document} and has a value between zero and one. A zero value of gamma indicates the absence of technical inefficiency among smallholder farmers, whereas a value of one implies that all deviations are due to technical inefficiency. Moreover, returns to scale, frontier yield, and the efficiency yield gap were calculated via the formula below.2.9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{R}\mathrm{T}\mathrm{S}=\sum\:_{\mathrm{i}=1}^{\mathrm{k}}{{\upbeta\:}}_{\mathrm{i}};\:{{\upbeta\:}}_{\mathrm{i}}=\frac{\partial\:\mathrm{l}\mathrm{n}{\mathrm{Y}}_{\mathrm{i}}}{\partial\:\mathrm{l}\mathrm{n}{\mathrm{X}}_{\mathrm{i}}}$$\end{document}2.10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\mathrm{Y}}_{\mathrm{i}}}^{\mathrm{*}}=\frac{{\mathrm{Y}}_{\mathrm{i}}}{{\mathrm{T}\mathrm{E}}_{\mathrm{i}}}$$\end{document}2.11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{y}\:{\mathrm{Y}}_{\mathrm{g}}={{\mathrm{Y}}_{\mathrm{i}}}^{\mathrm{*}}-\:{\mathrm{Y}}_{\mathrm{i}}$$\end{document} Where:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:{\mathrm{T}\mathrm{E}}_{\mathrm{i}}$$\end{document} is technical efficiency score; YHF is highest farmers yield; YW is water limited potential yield; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\mathrm{Y}}_{\mathrm{i}}}^{\mathrm{*}}$$\end{document} is frontier yield; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{Y}}_{\mathrm{i}}$$\end{document} is actual farm yield and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{T}\mathrm{E}}_{\mathrm{i}}$$\end{document} is technical efficient yield. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{E}\mathrm{f}\mathrm{f}.\:{\mathrm{Y}}_{\mathrm{g}}$$\end{document} is efficiency yield gap. The efficiency yield gap is the difference between the technically efficient yield or frontier yield (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\mathrm{Y}}_{\mathrm{i}}}^{\mathrm{*}}$$\end{document}) and actual yield (Ya). Multinomial endogenous switching regression model (MESR) The MESR model involves examining the impacts of adopting crop diversification, agroforestry, and adjusting planting dates on outcome variables (TE and efficiency yield gap). The nonadoption (A0C0Af0) of the CSA technology is represented by j = 0, whereas other combinations are represented by j = 1, 2, 3, 4… 7. Thus, we express each regime of equations as follows:2.12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left\{\begin{array}{c}\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{i}\mathrm{m}\mathrm{e}\:1:{\mathrm{w}}_{\mathrm{i}0}={\mathrm{X}}_{\mathrm{i}}{{\upbeta\:}}_{0}+{{\upmu\:}}_{\mathrm{i}0}\:\mathrm{i}\mathrm{f}\:\mathrm{I}=0\\\:\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{i}\mathrm{m}\mathrm{e}\:\mathrm{J}:\:{\mathrm{w}}_{\mathrm{i}\mathrm{j}}={\mathrm{X}}_{\mathrm{i}}{{\upbeta\:}}_{\mathrm{j}}+{{\upmu\:}}_{\mathrm{i}\mathrm{j}}\:\mathrm{i}\mathrm{f}\:\mathrm{I}=\mathrm{J}\end{array}\right\}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\:\mathrm{w}}_{\mathrm{i}\mathrm{j}\:}$$\end{document}refers to the outcome variable for the ith farmer when adopting CSA technology j; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{X}}_{\mathrm{i}}$$\end{document} indicates a vector of explanatory variables; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\upbeta\:}}_{1}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\upbeta\:}}_{\mathrm{j}}$$\end{document} are parameters to be estimated; and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\upmu\:}}_{\mathrm{i}1}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\upmu\:}}_{\mathrm{i}\mathrm{j}}$$\end{document} refer to disturbance terms. In the above equation, Xi addresses the observed selection bias issue. However, the disturbance components of the equation above are associated when farmers’ decisions to use CSA technology and outcome variables at the same time are influenced by the same unobserved factors. Selection bias problems arise if the correlation coefficients of the disturbance terms do not equal zero. Estimates that are biased and inconsistent may arise from failure to consider this problem. Estimates that are biased and inconsistent may arise from failure to consider this problem. Thus, to eliminate the unobserved selection bias, the selectivity-correction factors derived after estimating the following equation fit into the MESR model. By including the selectivity correction terms, the following can be rewritten as follows:2.13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left\{\begin{array}{c}\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{i}\mathrm{m}\mathrm{e}\:1:{\mathrm{w}}_{\mathrm{i}0}={\mathrm{X}}_{\mathrm{i}}{{\upbeta\:}}_{0}+{{\uplambda\:}}_{0}{{\upsigma\:}}_{0}+{{\upepsilon\:}}_{\mathrm{i}1}\:\mathrm{i}\mathrm{f}\:\mathrm{I}=0\\\:\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{i}\mathrm{m}\mathrm{e}\:\mathrm{J}:\:{\mathrm{w}}_{\mathrm{i}\mathrm{j}}={\mathrm{X}}_{\mathrm{i}}{{\upbeta\:}}_{\mathrm{j}}+{{\uplambda\:}}_{\mathrm{j}}{{\upsigma\:}}_{\mathrm{j}}+{{\upepsilon\:}}_{\mathrm{i}\mathrm{j}}\:\mathrm{i}\mathrm{f}\:\mathrm{I}=\mathrm{J}\end{array}\right\}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:{{\uplambda\:}}_{1\:}$$\end{document}and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\uplambda\:}}_{2}$$\end{document}refer to selectivity-correction terms and where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\upsigma\:}}_{1}\:$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\upsigma\:}}_{\mathrm{j}}$$\end{document} are the covariance between disturbance errors. Furthermore, J − 1 selectivity-correction terms are permitted by the multinomial selection framework, where one term corresponds to each possible combination of CSA technologies. Average treatment effects (ATT) Comparing the expected outcome variables of CSA technology adopters and nonadopters allows for the computation of the average treatment effect on the treated (ATT) in the third stage. The MESR model allows the computation of the counterfactual result, or what would have an outcome to CSA technology adopters if they had not embraced the technology, which allows for the computation of the ATT. The equations that follow are used to calculate the ATT estimations for the actual and counterfactual situations. CSA technology adopters with adoption (actual adoption observed):2.14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left\{\begin{array}{c}\mathrm{E}\left({\mathrm{W}}_{\mathrm{i}2}|\mathrm{I}=2\right)={\mathrm{X}}_{\mathrm{i}}{{\upbeta\:}}_{2}+{{\uplambda\:}}_{2}{{\upsigma\:}}_{2}\\\:\\\:\mathrm{E}\left({\mathrm{W}}_{\mathrm{i}\mathrm{j}}|\mathrm{I}=\mathrm{j}\right)={\mathrm{X}}_{\mathrm{i}}{{\upbeta\:}}_{\mathrm{j}}+{{\uplambda\:}}_{\mathrm{j}}{{\upsigma\:}}_{\mathrm{j}}\end{array}\right\}$$\end{document} The CSA technology adopters had they decided not to adopt (counterfactual):2.15\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left\{\begin{array}{c}\mathrm{E}\left({\mathrm{W}}_{\mathrm{i}0}|\mathrm{I}=2\right)={\mathrm{X}}_{\mathrm{i}}{{\upbeta\:}}_{0}+{{\uplambda\:}}_{1}{{\upsigma\:}}_{0}\\\:\\\:\mathrm{E}\left({\mathrm{W}}_{\mathrm{i}\mathrm{j}}|\mathrm{I}=\mathrm{j}\right)={\mathrm{X}}_{\mathrm{i}}{{\upbeta\:}}_{0}+{{\uplambda\:}}_{\mathrm{j}}{{\upsigma\:}}_{0}\end{array}\right\}$$\end{document} Finally, subtracting Eq. (3.7) from Eq. (3.8) gives the ATT. It is specified as2.16\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{A}\mathrm{T}\mathrm{T}=\mathrm{E}\left[\left({\mathrm{W}}_{\mathrm{i}2}|\mathrm{I}=2\right)-\left({\mathrm{W}}_{\mathrm{i}0}|\mathrm{I}=2\right)\right]={\mathrm{X}}_{\mathrm{i}}\left({{\upbeta\:}}_{2}-{{\upbeta\:}}_{0}\right)+{{\uplambda\:}}_{2}\left({{\upsigma\:}}_{2}-{{\upsigma\:}}_{0}\right)$$\end{document} Results and discussion The econometric model results are presented in this section. Descriptive statistics are also included. Descriptive results Tables 1 and 2 are list household attributes and CSA adoption rates. Out of 385 maize farmers, nonadopters (A0C0Af0) were 3.64%, while single-practice adopters were 3.90%–10.39%. Combined practices crop diversification with altered planting dates (A1C1Af0) and agroforestry with altered plant dates consisting of 20% and 7.27%, respectively. The most prevalent frequency package (43.38%) was the full combination of all three practices (A1C1Af1). Table 2 gives the eight CSA technology packages. The average values of nonadopter households (A0C0Af0) are used as a reference group to compare with the average values of adopter households (A1C0Af0, A0C1Af0, A0C0Af1, A1C1Af0, A1C0Af1, A0C1Af1, and A1C1Af1 packages). The average technical efficiency score is higher for adopters than for nonadopters. Furthermore, there are significant variations in the mean values of the outcome variables among the various adopters. The results indicated that all adopters had a lower efficiency yield gap than nonadopters did. Compared with farmers who accepted a single technology and nonadopters, farmers who adopted at least two types of CSA technologies had a lower efficiency yield gap. Table 1Alternative CSA package adoption and frequency.ChoiceMultinomial packagesAdjust plant dateCrop diversificationAgro-forestryFrequencyPercentageA0A1C0C1Af0Af10A0C0Af0✔✔✔143.641A1C0Af0✔✔✔153.902A0C1Af0✔✔✔4010.393A0C0Af1✔✔✔215.454A1C1Af0✔✔✔7720.005A1C0Af1✔⎫✔287.276A0C1Af1✔✔✔235.977A1C1Af1✔✔✔16743.38Note: Each element in the combination is a multinomial variable, and the subscripts represent 1 = adoption and 0 = nonadoption. Choice 0 represents those who do not adopt any of these technologies (A0C0Af0); 1 is for adjusting plant date adopters (A1C0Af0); 2 is for crop diversification adopters (A0C1Af0); 3 is for agroforestry adopters (A0C0Af1); 4 is for adjusting plant date and crop diversification adopters (A1C1Af0); 5 is for adjusting plant date and agroforestry adopters (A1C0Af1); 6 is for crop diversification and agroforestry adopters (A0C1Af1); and 7 is for adjusting plant date, crop diversification, and agroforestry adopters (A1C1Af1). Table 1Alternative CSA package adoption and frequency.ChoiceMultinomial packagesAdjust plant dateCrop diversificationAgro-forestryFrequencyPercentageA0A1C0C1Af0Af10A0C0Af0✔✔✔143.641A1C0Af0✔✔✔153.902A0C1Af0✔✔✔4010.393A0C0Af1✔✔✔215.454A1C1Af0✔✔✔7720.005A1C0Af1✔⎫✔287.276A0C1Af1✔✔✔235.977A1C1Af1✔✔✔16743.38Note: Each element in the combination is a multinomial variable, and the subscripts represent 1 = adoption and 0 = nonadoption. Choice 0 represents those who do not adopt any of these technologies (A0C0Af0); 1 is for adjusting plant date adopters (A1C0Af0); 2 is for crop diversification adopters (A0C1Af0); 3 is for agroforestry adopters (A0C0Af1); 4 is for adjusting plant date and crop diversification adopters (A1C1Af0); 5 is for adjusting plant date and agroforestry adopters (A1C0Af1); 6 is for crop diversification and agroforestry adopters (A0C1Af1); and 7 is for adjusting plant date, crop diversification, and agroforestry adopters (A1C1Af1). Alternative CSA package adoption and frequency. Note: Each element in the combination is a multinomial variable, and the subscripts represent 1 = adoption and 0 = nonadoption. Choice 0 represents those who do not adopt any of these technologies (A0C0Af0); 1 is for adjusting plant date adopters (A1C0Af0); 2 is for crop diversification adopters (A0C1Af0); 3 is for agroforestry adopters (A0C0Af1); 4 is for adjusting plant date and crop diversification adopters (A1C1Af0); 5 is for adjusting plant date and agroforestry adopters (A1C0Af1); 6 is for crop diversification and agroforestry adopters (A0C1Af1); and 7 is for adjusting plant date, crop diversification, and agroforestry adopters (A1C1Af1). Table 2Descriptive results for the outcome variables used in the regression.Outcome variableMean(Std Dev)A0C0Af0A1C0Af0A0C0Af1A0C1Af0A1C0Af1A1C1Af0A0C1Af1A1C1Af1OverallTechnical efficiency0.34(0.12)0.46***(0.02)0.56***(0.01)0.51***(0,02)0.71***(0.01)0.64***(0.03)0.74***(0.01)0.82***(0.04)0.69(0.14)Efficiency Yield gap4.63(0.49)3.36***(0.14)3.69***(0.29)2.79***(0.22)1.83***(0.03)2.12***(0.15)1.67***(0.06)1.06***(0.32)1.91(1) Table 2Descriptive results for the outcome variables used in the regression.Outcome variableMean(Std Dev)A0C0Af0A1C0Af0A0C0Af1A0C1Af0A1C0Af1A1C1Af0A0C1Af1A1C1Af1OverallTechnical efficiency0.34(0.12)0.46***(0.02)0.56***(0.01)0.51***(0,02)0.71***(0.01)0.64***(0.03)0.74***(0.01)0.82***(0.04)0.69(0.14)Efficiency Yield gap4.63(0.49)3.36***(0.14)3.69***(0.29)2.79***(0.22)1.83***(0.03)2.12***(0.15)1.67***(0.06)1.06***(0.32)1.91(1) Descriptive results for the outcome variables used in the regression. 0.34 (0.12) 0.46*** (0.02) 0.56*** (0.01) 0.51*** (0,02) 0.71*** (0.01) 0.64*** (0.03) 0.74*** (0.01) 0.82*** (0.04) 0.69 (0.14) 4.63 (0.49) 3.36*** (0.14) 3.69*** (0.29) 2.79*** (0.22) 1.83*** (0.03) 2.12*** (0.15) 1.67*** (0.06) 1.06*** (0.32) 1.91 (1) The average efficiency yield gap was 1.91 t/ha (see Table 2). This shows that with present resources and technology, the average farmer could increase current yield by 1.9 t/ha if inputs were used efficiently. Thus, differences in crop management techniques, such as methods of soil preparation, nutrient management, pest and disease control, and labor-intensive tasks, constitute the root cause of the technical efficiency yield gap71. Impact of CSA practice on the technical efficiency score Table 3 shows the average treatment effects from the multinomial endogenous switching regression model on the impact of adopting various CSA practices on the technical efficiency score in both the actual and counterfactual scenarios. The average treatment effects on the treated are the difference in expected output between those who have chosen to adopt (adopters) and the efficiency and yield gap of farmers who did not adopt the CSA practices they adopted. The findings indicate significant variations between the actual and counterfactual outcomes, implying that adopters of any CSA practices have greater average technical efficiency scores than they would have if they had not adopted. For example, implementing crop diversification with agroforestry (A0C1Af1) jointly considerably increases the technical efficiency score by 0.25 in comparison with nonadoption (A0C0Af0). Compared with nonadopters, the adoption of adjusted plant dates enhances TE by 25%. This implies that the mean TE scores increased from 47% to 71%. The results also show that, relative to nonadoption (A0C0Af0), adopting an adjusted plant date only (A1C0Af0), crop diversification only (A0C1Af0), and agroforestry only (A0C0Af1) significantly increase the technical efficiency score by 0.25, 0.57, and 0.54, respectively. The results reveal that the highest technical efficiency score (0.57) is obtained when farmers adopt crop diversification only (A0C1Af0), followed by agroforestry adoption (0.54) and a combination of adjusting plant date and crop diversification adoption (0.38). The outcomes of this investigation align with the findings of72, who demonstrated that, compared with nonadoption, the use of multiple strategies achieves the highest TE. The negative BH1 (except A1C1Af0, all are negative) indicates that nonadopters would have had a higher technical efficiency score than those who had adopted if they had decided to adopt. The transitional heterogeneity effect (TH) for the adoption of CSA (A1C0Af0, A0C1Af0, and A1C0Af1) practices is negative and significant. This also means that the likely impact of adoption of CSA practices would be much greater in the case of the non-adopting household compared to the adopting household if both groups were to adopt. This study complements the findings of73,74, who discovered that adopters had greater levels of TE than nonadopters did. This suggests that farmers who use technology are technically more efficient than those who do not75. suggested that organic farms would have 10.7% points less TE if they had not adopted the organic farming strategy, in contrast to conventional farms, which can increase their TE by 8.6% points by adopting organic farming76. also proposed that adopters of improved rice varieties have 24% greater technical efficiency than nonadopters do. The expansion of climate change adaptation has led to increased farmer efficiency, according to77–79, and80. Table 3Conditional average treatment, transitional and heterogeneity effects of the technical efficiency score outcome.TE scoreCSA PackagesDecision stageAverage TreatmenteffectTo adoptNot to adoptFarm households That AdoptedTreatedA1C0Af00.71(0.03)0.47(0.05)0.25***(0.06)A0C1Af00.58(0.08)0.06(0.05)0.57***(0.09)A0C0Af10.70(0.01)0.15(0.07)0.54***(0.06)A1C1Af00.72(0.01)0.35(0.04)0.38***(0.04)A1C0Af10.78(0.01)0.46(0.03)0.31***(0.04)A0C1Af10.77(0.01)0.53(0.05)0.25***(0.04)A1C1Af10.73(0.002)0.53(0.02)0.20***(0.02)UntreatedThat didn’tAdoptedA1C0Af01.29(0.25)0.71(0.04)0.58**(0.26)A0C1Af01.29(0.25)0.71(0.04)0.58**(0.26)A0C0Af10.78(0.08)0.71(0.04)0.07(0.09)A1C1Af00.64(0.04)0.71(0.04)−0.07(0.05)A1C0Af10.81(0.05)0.71(0.04)0.10(0.06)A0C1Af10.81(0.05)0.71(0.04)0.10(0.06)A1C1Af10.80(0.01)0.71(0.04)0.09**(0.04)BH1BH0THTransitional andHeterogeneityEffectsA1C0Af0−0.58**(0.25)−0.25***(0.07)−0.33(0.25)A0C1Af0−0.71***(0.20)−0.70***(0.09)−0.01(0.2)A0C0Af1−0.08(0.07)−0.55***(0.09)0.48***(0.10)A1C1Af00.08***(0.02)−0.36***(0.10)0.44***(0.11)A1C0Af1−0.44***(0.09)−0.25***(0.06)−0.19*(0.10)A0C1Af1−0.04(0.04)−0.18***(0.07)0.14**(0.07)A1C1Af1−0.07***(0.01)−0.17**(0.07)0.10(0.07)*, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. Table 3Conditional average treatment, transitional and heterogeneity effects of the technical efficiency score outcome.TE scoreCSA PackagesDecision stageAverage TreatmenteffectTo adoptNot to adoptFarm households That AdoptedTreatedA1C0Af00.71(0.03)0.47(0.05)0.25***(0.06)A0C1Af00.58(0.08)0.06(0.05)0.57***(0.09)A0C0Af10.70(0.01)0.15(0.07)0.54***(0.06)A1C1Af00.72(0.01)0.35(0.04)0.38***(0.04)A1C0Af10.78(0.01)0.46(0.03)0.31***(0.04)A0C1Af10.77(0.01)0.53(0.05)0.25***(0.04)A1C1Af10.73(0.002)0.53(0.02)0.20***(0.02)UntreatedThat didn’tAdoptedA1C0Af01.29(0.25)0.71(0.04)0.58**(0.26)A0C1Af01.29(0.25)0.71(0.04)0.58**(0.26)A0C0Af10.78(0.08)0.71(0.04)0.07(0.09)A1C1Af00.64(0.04)0.71(0.04)−0.07(0.05)A1C0Af10.81(0.05)0.71(0.04)0.10(0.06)A0C1Af10.81(0.05)0.71(0.04)0.10(0.06)A1C1Af10.80(0.01)0.71(0.04)0.09**(0.04)BH1BH0THTransitional andHeterogeneityEffectsA1C0Af0−0.58**(0.25)−0.25***(0.07)−0.33(0.25)A0C1Af0−0.71***(0.20)−0.70***(0.09)−0.01(0.2)A0C0Af1−0.08(0.07)−0.55***(0.09)0.48***(0.10)A1C1Af00.08***(0.02)−0.36***(0.10)0.44***(0.11)A1C0Af1−0.44***(0.09)−0.25***(0.06)−0.19*(0.10)A0C1Af1−0.04(0.04)−0.18***(0.07)0.14**(0.07)A1C1Af1−0.07***(0.01)−0.17**(0.07)0.10(0.07)*, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. Conditional average treatment, transitional and heterogeneity effects of the technical efficiency score outcome. That didn’t Adopted Transitional and Heterogeneity Effects *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. Impact of the adoption of CSA practices on the yield gap efficiency Table 4 presents the average treatment effects of adopting different CSA practices on the efficiency yield gap under actual and counterfactual scenarios from the multinomial endogenous switching regression model on the treated (ATT). The findings indicate significant variations between the actual and counterfactual outcomes, implying that adopters of any CSA practices have smaller average efficiency yield gaps in t/ha than they would have if they had not adopted. For example, implementing crop diversification with agroforestry (A0C1Af1) decreases the efficiency yield gap by 3.67 t/ha in comparison to nonadoption (A0C0Af0). The results show that, relative to nonadoption (A0C0Af0), adopting an adjusted plant date only (A1C0Af0) or agroforestry only (A0C0Af1) significantly decreases the efficiency yield gap by 2.52 t/ha and 1.16 t/ha, respectively. Furthermore, the study revealed that a combination of multiple CSA practices (A1C1Af1) could result in a lower efficiency yield gap (t/ha) than nonadoption CSA practices. The result reveals that the lowest efficiency yield gap (−4.62 t/ha) is obtained when farmers adopt a combination of full technology adopters (A1C1Af1), a combination of adjusting plant date and agroforestry adoption (−3.31 t/ha), and a combination of adjusting plant date and crop diversification adoption (−3.12 t/ha). The negative BH1 values (all except A0C1Af0 and A0C0Af1) indicate that nonadopters would have a lower efficiency yield gap (t/ha) than those who had adopted if they had decided to adopt. The transitional heterogeneity effect (TH) for the adoption of CSA practices is negative and significant (except for A0C1Af0 and A0C0Af1), which implies that the effect of adoption would be a significantly lower efficiency maize yield gap for the farm households who had not adopted relative to their counterfactual. This also implies that, if both groups adopt, the non-adopting household is likely to experience much greater impact from the adoption of CSA practices than would the adopting household. This study result agrees with that of81; existing yields could be more than doubled by closing efficiency yield gaps through better crop management (i.e., adjusted planting time). Climate change has been shown to have a negative impact on cereal crop yields, with studies by82 showing a 60% decrease in maize yields. On the other hand83, reported that crop rotation strategies significantly increased the rice yield per acre, ranging from 1.39 to 2.08 quintals. Furthermore84, reported that N fertilizer enhanced yield when combined with adjusted plant date practices. Table 4Conditional average treatment and transitional and heterogeneity effects of the yield gap outcome.Efficiency Yield gapCSA PackagesDecision stageAverage TreatmenteffectTo adoptNot to adoptFarm households That AdoptedTreatedA1C0Af03.36(0.04)5.88(0.22)−2.52***(0.22)A0C1Af03.75(0.06)3.67(0.17)0.08(0.18)A0C0Af13.70(0.06)4.85(0.26)−1.16***(0.28)A1C1Af02.12(0.01)5.25(0.17)−3.12***(0.17)A1C0Af11.83(0.004)5.14(0.23)−3.31***(0.23)A0C1Af11.67(0.01)5.35(0.30)−3.67***(0.30)A1C1Af11.06(0.01)5.68(0.08)−4.62***(0.08)UntreatedThat didn’tAdoptedA1C0Af03.74(0.07)4.63(0.13)−0.89***(0.15)A0C1Af03.74(0.07)4.63(0.13)−0.89***(0.15)A0C0Af10.90(0.4)4.63(0.13)−3.73***(0.42)A1C1Af02.19(0.02)4.63(0.13)−2.43***(0.13)A1C0Af11.95(0.03)4.63(0.13)−2.68***(0.13)A0C1Af11.69(0.02)4.63(0.13)−2.94***(0.13)A1C1Af11.07(0.04)4.63(0.13)−3.56***(0.14)BH1BH0THTransitional andHeterogeneityEffectsA1C0Af0−0.38***(0.08)1.25***(0.26)−1.63***(0.30)A0C1Af00.01(0.10)−0.95***(0.30)0.97***(0.33)A0C0Af12.80***(0.33)0.23(0.35)2.57***(0.47)A1C1Af0−0.07***(0.02)0.62(0.41)−0.69*(0.41)A1C0Af1−0.12***(0.02)0.51(0.35)−0.63*(0.35)A0C1Af1−0.01(0.02)0.72*(0.40)−0.73*(0.40)A1C1Af1−0.01(0.03)1.05***(0.28)−1.06***(0.28)*, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. Table 4Conditional average treatment and transitional and heterogeneity effects of the yield gap outcome.Efficiency Yield gapCSA PackagesDecision stageAverage TreatmenteffectTo adoptNot to adoptFarm households That AdoptedTreatedA1C0Af03.36(0.04)5.88(0.22)−2.52***(0.22)A0C1Af03.75(0.06)3.67(0.17)0.08(0.18)A0C0Af13.70(0.06)4.85(0.26)−1.16***(0.28)A1C1Af02.12(0.01)5.25(0.17)−3.12***(0.17)A1C0Af11.83(0.004)5.14(0.23)−3.31***(0.23)A0C1Af11.67(0.01)5.35(0.30)−3.67***(0.30)A1C1Af11.06(0.01)5.68(0.08)−4.62***(0.08)UntreatedThat didn’tAdoptedA1C0Af03.74(0.07)4.63(0.13)−0.89***(0.15)A0C1Af03.74(0.07)4.63(0.13)−0.89***(0.15)A0C0Af10.90(0.4)4.63(0.13)−3.73***(0.42)A1C1Af02.19(0.02)4.63(0.13)−2.43***(0.13)A1C0Af11.95(0.03)4.63(0.13)−2.68***(0.13)A0C1Af11.69(0.02)4.63(0.13)−2.94***(0.13)A1C1Af11.07(0.04)4.63(0.13)−3.56***(0.14)BH1BH0THTransitional andHeterogeneityEffectsA1C0Af0−0.38***(0.08)1.25***(0.26)−1.63***(0.30)A0C1Af00.01(0.10)−0.95***(0.30)0.97***(0.33)A0C0Af12.80***(0.33)0.23(0.35)2.57***(0.47)A1C1Af0−0.07***(0.02)0.62(0.41)−0.69*(0.41)A1C0Af1−0.12***(0.02)0.51(0.35)−0.63*(0.35)A0C1Af1−0.01(0.02)0.72*(0.40)−0.73*(0.40)A1C1Af1−0.01(0.03)1.05***(0.28)−1.06***(0.28)*, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. Conditional average treatment and transitional and heterogeneity effects of the yield gap outcome. That didn’t Adopted Transitional and Heterogeneity Effects *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. Conclusion and policy implications Current climatic change and variability, including natural disasters, resource degradation, and effects on food security, the availability of water, poverty alleviation, and sustainable development initiatives, have created serious challenges in the western region of Ethiopia. Therefore, to conduct the necessary policy intervention to improve yields of maize, understanding the role that the CSA has in bridging the efficiency yield gaps is crucial. By thoroughly examining the effects of CSA on efficiency yield gaps and offering potential policy solutions, the current study is expected to fill a research gap. The findings indicate that adopters of any CSA practices have a greater average technical efficiency score and a smaller average efficiency yield gap in t/ha than they would have if they had not adopted them. Furthermore, the study revealed that a combination of CSA practices could result in a higher technical efficiency score and a lower efficiency yield gap (t/ha) than nonadopters. The results also show that, relative to nonadoption, adopting an adjusted plant date only, crop diversification only, and agroforestry significantly increase technical efficiency by 25%, 57%, and 54%, respectively. The results also show that, relative to nonadoption, adopting an adjusted plant date and agroforestry significantly decreases the efficiency yield gap by 2.52 t/ha and 1.16 t/ha, respectively. The study recommends that policies including CSA practices take smallholder farmer efficiency into account, as CSA strategies are most effective when farmers are sufficiently efficient in producing crops. A beneficial strategy to lessen the adverse impacts of climate change is to use multiple CSA techniques rather than a single technique. Furthermore, when combined CSAs are implemented, sufficient funds, incentives, support for technical help, and increased community involvement are advised.