1 Introduction

The economy and livelihoods of societies in Ethiopia depend heavily on agriculture [1]. However, agriculture is characterized as subsistent and dominated by smallholder farming system [2] while among the total area, 95% are cultivated by smallholder farmer [3]. It plays a vital role in the country’s gross domestic product (34%), foreign exchange earnings 75%), employment opportunities (73%), and provision of raw materials (70%) for agro-processing industries and a source of supply at market [4]. Ethiopia is the largest wheat producer in sub-Saharan Africa countries [5, 6] and about 5.60 million metric tons of wheat is produced on average in the country [7] with the productivity of 33.64 qt/ha [8]. It is seasonally produced during rainy time June to October [9] and largely produced in the central, southeastern, and northwestern highland areas of the country [10]. The country is planning and implementing strategies to improve wheat productivity [11] although its yield is comparatively low (2.4 tons per hectare) while the global standard wheat production average is 3.4 tons per hectare [12]. Low wheat production is caused by the limited use of production inputs, such as the use of improved seeds, fertilizers, pesticides, lack of access to technology, subsistent production, drought, and weak market linkages [2, 13, 14].

In low-income countries, such as Ethiopia, where the majority of the population derives their livelihood from agriculture, the adoption of improved agricultural technology practices substantially improves agricultural production and predominantly contributes to agricultural development [15, 16]. However, the success of technology adoption relies on the availability and affordability of technologies, expectations of their profitability, and associated risks [17]. The extent of agricultural technology adoption is impacted by factors such as the attitude and perceptions of adopters, cost of adoption, and factors related to technological and geographical factors [18] while controlling these factors facilitates the diffusion of technologies [19]. However, the effect of agricultural technology adoption and its diffusion becomes more valuable when adopted in combination packages [20]. The adoption of such technology has a sustainable effect on agricultural productivity when the full extents of the new and existing technologies are continuously adopted over time [21, 22]. In developing countries such as Ethiopia, although agricultural technology adoption is considered as an instant option by various development actors and policymakers to improve agricultural productivity and sustain household well-being [23], adoption rates of important agricultural technologies remain low in the region [23, 24] causing low agricultural productivity and household poverty [20].

The low adoption rate of these technologies is caused predominantly by adopters’ internal and external production constraints, such as lack of credit, problems related with agricultural production, market problems, farmers’ behavior, technology constraints, and lack of advisory services [22]. Although these technology package adoption constraints force farmers to adopt a single package or a few packages, the interdependency of the technologies and their likely effect on each other enable farmers to jointly select technology packages [25]. Moreover, the adoption of modern agricultural technology in recommended packages further improves the highest possible yields and economic profits [26]. More probably, the agricultural technology packages consist of the application of planting in rows, improved seed varieties, weeding rates, applications of chemicals, and fertilizers. Making a proactive decision on technology package selection helps farmers to minimize challenges and apply recommended technologies to achieve the desired outcome of technology adoption, such as improving productivity [27, 28].

In Ethiopia, various policies aimed at improving agricultural productivity to reduce smallholder farmers’ poverty and improve their self-sustaining food have been implemented [24]. By realizing the effect of agricultural technology adoption on farmers’ livelihoods, many institutions, such as agricultural research institutions, universities, and vocational colleges have tried to develop and disseminate various agricultural technologies [29] but the expected outcome is not realized in the country. Further, to improve the productivity of agriculture, Ethiopia has organized agricultural extension package implementation in rural areas to disseminate and adopt agricultural inputs, such as improved seed variety, application of fertilizers, chemicals, management practices, and other related agricultural technologies [30, 31]. Additionally, many efforts have been made to adjust diversity in the biophysical and socioeconomic adoption of agricultural technology packages though not achieved the expected outcome in the country [32]. Despite these all efforts, the country’s level of adoption of new agricultural technologies is low. Previous studies on agricultural technology adoption, such as refs. [15, 24], discussed the low adoption rate of agricultural technologies rather than focusing on why farmers fail to adopt recommended technologies and the determinant factors behind these issues.

Like other cereals, improvement in wheat productivity needs application of technology packages such as improved varieties, seeding rates, row planting, better irrigation, weed management, and use of recommended fertilizer rates to be implemented and disseminated to farmers [33, 34]. However in Ethiopia, despite implementing these technology packages and national strategies for wheat production, a huge wheat productivity gap exists and the domestic demand for wheat has forced the country to import wheat [9, 35]. These were caused by incomplete technology package adoption. Selecting and relying on a single technology adoption lacks better outcomes, as technology packages are interdependent [25]. However, the adoption of all recommended wheat production technology packages realizes the expected aim of improving wheat production and productivity, and meeting domestic wheat demand.

Most studies point out the consequences of lack of access to improved wheat varieties, application of fertilizers, and pesticide on wheat production, and fail to discuss the effect of other technology packages [5, 6, 36,37,38,39]. Moreover, none of these studies identified the extent of adoption of these technology packages and the potential effect of the lack of full application of technology in packages on the extent of production and productivity.

In the study area, lack of access to quality improved wheat seed at the right quantity, at the right price, and its availability for smallholder farmers is one of the most challenging issues that causes inappropriate seed systems and contributes to low wheat productivity. Unfair wheat seed prices and the presence of illegal traders are preventing farmers from getting access to available wheat seed. As a result of all these challenges, farmers’ adoption rates of recommended improved seed are low. In addition, wheat row planting is not well practiced, and most farmers sow seeds by broadcasting. The low level of wheat row planting adoption rate is mostly related to farmers’ lack of awareness and area of land under broadcasting is greater than the area under row planting. The price of inorganic fertilizers is continuously increasing, which farmers are unable to afford it in the study area. The availability of fertilizer at the right time with the right quantity is also the most influential problem, and farmers face the constraints of the adoption of recommended fertilizers for a given size of land, and their adoption level is too low. Therefore, to maximize smallholder farmers’ expected wheat production level per given area of land, this study looks at the importance of recommended wheat production input combinations and their sowing methods. Overall, in the study area, smallholders’ adoption level of these technology packages is found to be low, caused by a lack of access to these technologies, a lack of awareness, and lack of supporting services. However, adoption of all recommended wheat technology package adoption could improve smallholder farmers’ wheat productivity, and overall household wellbeing.

Similarly, empirical studies on wheat technology adoption in Ethiopia, such as refs. [40,41,42,43,44,45,46,47,48,49] focused on the adoption of a single or few wheat production technology packages and failed to select multiple wheat production technology packages that could improve the level of production. These factors caused farmers to produce below the estimated amount of production. There are scant empirical literature on the demographic and socio-economic factors of adoption of wheat production technology packages such as improved wheat variety, wheat row planting, and application of recommended agro-chemicals. In addition, no empirical study conducted to identify the level of wheat production technology packages adoption and to identify factors associated with farmers’ low adoption level by using these wheat technology packages. Moreover, it is very difficult to derive actual households’ adoption status from agricultural technology adoption undertaken based on a single or few technology packages, and this will fail to be an input for devising agricultural policies. Hence, these literature gaps initiated this study to be conducted from the perspectives of assessing the adoption of recommended wheat technology packages. Furthermore, adoption of recommended wheat technology packages and understanding its associated factors could enable to formulate concrete agricultural policies and strategies and researchers to further improve households’ recommended wheat technology package adoption.

2 Literature reviews

Factors of households’ technology adoption intensity can be theoretically understood depending on the behaviors and perceptions of farmers, nature, categories, spread of diffusion, and objectives of the existing technology/s. Moreover, the issues related to adoption intensity of agricultural technologies are determined by the time variation, farmers’ willingness and ability to adopt new agricultural technologies, efforts to create awareness about technology adoption, its trial procedures, and decision making on adoption of all recommended technology packages that provide the expected outcome. Specifically, farmers’ attitudes, wealth, education, landholdings, and characteristics of the technology are highly linked to adoption intensity. In refs. [50,51,52,53] found that farmers’ behavior, nature of the technology, objectives of the technology, and diffusion of technology adoption could directly or indirectly influence the extent of technology adoption. In addition, the success rate of technology adoption depends on time, integration of a new technology into existing practices, and adoption of technologies in combinations [54,55,56,57,58]. Specifically, farmers’ demographic and socioeconomic characteristics, resource endowment, and farm attributes are factors that influence the adoption intensity of agricultural technologies. These types of reviews were included in this study to further understand the existing literature gaps.

The empirical research findings of refs. [49, 59,60,61,62,63,64,65,66,67,68] revealed that the demographic characteristics of households, such as household age, family size, and level of education, significantly influenced the adoption of agricultural technologies. The adoption and intensity of agricultural commodities are influenced by extension contact, access to credit, agricultural training, farmers’ cooperative membership, demonstration participation, and seed multiplication membership [32, 49, 59,60,61, 63, 64, 67, 69,70,71]. Households’ distance from their residence to the main market, input market, main road, training centers, and farm areas are determinant factors affecting the adoption of agricultural technology packages [67, 69]. Resource ownership such as landholding, livestock ownership, and number of oxen owned were important factors in determining the level of households’ agricultural technology adoption [49, 64, 67, 71, 72] while income from farm and non-farm were influential factors in determining adoption of agricultural technologies [32, 65, 67]. Moreover, the location of farmers (region) had an impact on the adoption of agricultural technologies [32, 68], owning communication media such as radio and mobile also were influential factors of agricultural technology adoption [71] and households’ perception about yield significantly affected agricultural technology adoption [65, 69].

In ref. [73] studied adoption of improved bread wheat varieties in Ethiopia by employing binary probit and identified that educational level, family labor, oxen ownership, training access, membership in cooperatives, credit access and age are adoption associated factors. Similarly, ref. [40] discussed adoption of a single technology package (wheat row planting) by using a binary logistic regression model and revealed that age, education, farming experience, labor, annual income and extension service are associated factors. Both these empirical studies focused on adoption decision of a single technology and didn’t consider adoption of recommended technology packages and its intensity of adoption which could cause incomplete technology adoption and low wheat productivity. According to refs. [74, 75], the decision to adopt agricultural production technology is undertaken by considering different factors, such as household characteristics (e.g., age, sex, family member size), household socioeconomic factors (e.g., educational level, farm size, livestock ownership, income, technology adoption costs), institutional factors (e.g., access to credit, extension workers’ contacts, membership to social services, distance from the nearest service providers), and technological factors (e.g., information about access to new technologies).

The research findings of ref. [76] revealed that the adoption of agricultural production package technologies in Ethiopia is limited, although it can contribute many advantages in improving agricultural production, generating income and food security. A combination of different production input technologies can be adopted jointly or disaggregated. Agricultural technologies are either limited to a single, a few, or a combination of few technologies. As a result, the adoption of agricultural production technologies became low and did not expand in the area allocated for agricultural production. Double-hurdle models were used to analyze three rounds of panel data collected from smallholder farms for a total sample of 1269 households. The results indicate that agricultural technology adoption and its intensity of technology adoption are affected by demographic, weather, and market factors. According to the results, landholding size and access to irrigation had a positive effect on adoption, and smallholders who adopted new crop technology showed a small increase in adoption over several years. In ref. [42] discussed that the low adoption rate of high-yielding wheat varieties causes low wheat productivity, though the study was conducted by using a single technology package, which could be a factor in low wheat productivity. The authors employed double hurdle model and identified that seed availability, row planting, distance to the cooperative, and farmland allotted determined the adoption intensity of the improved wheat variety, while its adoption decision was influenced by associated factors such as farming experience, distance to the cooperative, renting a tractor and combine harvester, urea application, and wheat sale income.

3 Research methodology

3.1 Description of the study area

This study was conducted in the Horo Guduru Wollega zone of the western Oromia region, Ethiopia. The area lies between latitude 9° 10′ N and 9° 50′ N and longitude 36° 00′ E and 36° 50′ E direction. The zone was purposively selected from among zones located in Western Oromia region, based on its potential for wheat production. Overall, the zone is well known for wheat production, and wheat is produced in all districts of the zone, which helps many smallholder farmers improve their daily livelihood by using wheat production for personal consumption and as a source of income. However, the extent of wheat production in the study area is still dominated by backward production systems. Farmers are not advanced with modern agricultural technologies, and do not efficiently utilize technologies through adoption. Farmers face problems in the supply of agricultural inputs during production seasons and are forced to purchase inputs at high costs from informal traders. In addition to wheat production, the zone is advanced with the production of oil crops, other cereals, fruits, vegetables, pulses, and so on. Rearing local and hybrid livestock is widely undertaken in the study area. The study area is characterized by different agro-ecologies, such as lowlands, midlands, and highlands. This zone has also suitable climatic conditions for wheat production, although the extent of production is not as expected. Its average annual temperature is 22.1 °C; with 13 °C and 30 °C minimum and maximum temperatures, respectively while it’s annual rainfall ranges from 1000 to 2400 mm [77]. Three districts, Abe Dongoro, Ababo Guduru, and Horo, were purposively selected from each agro-ecology found in the zone. Two kebeles were selected from each district, totaling six kebeles. Accordingly, Wirtu Senxa and Tulu Moti kebeles from the Abe Dongoro district, Ilamu Molale and Loya Malole kebeles from the Ababo Guduru district, and Laku Igu and Gitilo Dale kebeles from the Horo district were selected (Fig. 1).

Fig. 1
figure 1

Source: Adapted from Ethiopia Map

Administrative map of the study area.

Full size image

3.2 Data collection

To collect the data from different sources, different procedures were followed. The study used qualitative and quantitative data collected from different sources using semi-structured questionnaire. To collect the overall primary data, different activities, such as technology package identification, discussion with agricultural experts, site selection, preliminary activities through visiting specific study areas, and collecting data, were undertaken at different times. Accordingly, to collect the required data, enumerators were trained on methods of data collection and how to approach to respondents. A cross-sectional household survey method was employed to collect quantitative data from randomly selected farm households in the study areas during 2022/2023 agricultural production season. To improve the validity of the quantitative data, the study followed appropriate methodologies and designs. Questionnaires were translated to the farmers’ mother tongue (from English to Afan Oromo) to avoid any misunderstandings of the questions and enable them to fully participate in the interview. Finally, the expected data were successfully collected from head of the household (male or female) without any delays or incompleteness. Moreover, the authors checked the collected data for errors, incompleteness, and inconsistencies/discrepancies and no problem observed. To understand the wheat production technology adoption of smallholder farmers in the study area and to collect qualitative data, an oral interview was conducted with each district’s agricultural offices (agricultural extension workers), model farmers, and peasant association administrators. Additionally, members of focus group discussions (FGD) who were assumed to have in-depth knowledge about the purpose of this study and support the integrity of the results of the study were also interviewed. Using trained enumerators who speak the local language, both qualitative and quantitative data were collected from the sampled households through a survey. Secondary data were also collected from the agricultural office of the zone and from selected districts, Central Statistical Agencies (CSA), reports, and websites.

3.3 Sampling techniques and sample size

A combination of purposive and probability sampling techniques was employed to select farm households from wheat producers in the study area. First, the Horo Guduru Wollega zone is purposively selected depending on the potential of wheat production and the extent of application of resource utilization. The zone was then stratified into lowland, midland, and highland zones based on its agro-ecology. One (1) potential district was purposively selected from each stratum. At the next stage, the list of all selected wheat-producing kebeles was taken from each district’s agricultural offices and two (2) kebeles (to compare differences among groups) were randomly selected from each selected district. Finally, a list of wheat-growing farmers was obtained from the development agent offices of each kebele. Then, sample farm households were randomly selected from the list of wheat producer farmers. The ref. [78] formula was used for sample size determination to compare differences among groups in each agro-ecology.   The allocation of sample size to each kebele was undertaken proportional to the household size of each kebele within their respective districts (Table 1).

Table 1 Sample size of selected districts and kebeles.
Full size table
$$\text{n}= \frac{{\text{Z}}^{2}\text{pqN}}{{\text{e}}^{2}\left(\text{N}-1\right)+{\text{Z}}^{2}\text{pq}}$$
(1)

3.4 Estimation of econometric model (adoption index)

In many adoption studies, different econometric models such as double hurdle (DH) model [42, 75, 79] Heckpoisson regression model [80], tobit model [81, 82], a two-limit tobit model [32, 67] have been employed to analyze and identify the factors influencing the intensity (level) of agricultural technology adoption. The selection of an appropriate econometric model depends on the nature of the data, objectives/s, and type of dependent variable of the study [83]. According to ref. [84], to analyze adoption intensity and factors affecting adoption decisions, the DH model is preferably selected over other models when farmers’ access to agricultural inputs is constrained. For example, [69, 71, 79] used the DH model to analyze factors influencing agricultural technology adoption. It is not possible to employ the Ordinary Least Square (OLS) model for a categorical or qualitative analysis, as the parameter estimates obtained become inefficient and heteroscedastic problems will occur [85].

If the dependent variable bears continuous values below and above the limits (0 and 1), simultaneously censored from above and below, a two-limit Tobit model (censored regression model) can be used [86, 87]. In this study, the dependent variable was computed from different wheat production technology packages (indexed) commonly adopted by farmers. It bears values between 0 and 1, or outside of the ranges. Therefore, this study employed a two-limit Tobit model to analyze the determinants of wheat production technology packages adoption. Some authors, such as refs. [32, 67], also employed a two-limit Tobit model to assess the factors affecting the adoption of technology packages.

The index function of the general formulation of the two-limit Tobit model is

$${\text{Y*}} = \text{X}^{\prime}{\text{B}} + {\text{e}}$$
(2)
$${Y}_{i}=\left\{\begin{array}{c}0 if {Y}_{i}^{*}\le 0\\ {Y}_{i}^{*}if 0<{Y}_{i}^{*}<1\\ 1 if {Y}_{i}^{*}>1\end{array}\right\}$$
(3)

where Y* is the limited dependent variable, which is the intensity of adoption of wheat production technologies, X’ are explanatory variables, and the latter two equations represent a censored distribution of the data. The expected value of Yi as a function of a set of explanatory variables (X) weighted by the probability that Yi > 0 can be estimated using the Tobit model [86].

$${\text{E}}\left( {\text{Y}} \right) = {\text{X}}\upbeta {\text{F}}\left( {\text{z}} \right)+ \upsigma {\text{f}}\left( {\text{z}} \right){\text{and z}} = {\text{X}}\upbeta/\upsigma$$
(4)

where F(z) is the cumulative normal distribution of z, f(z) is the derivative of the normal curve), z is the score of the area under the normal curve, and s is the standard error of the error term. Depending on wheat production technology packages, such as improved wheat varieties, wheat row planting, and application of recommended fertilizer in the study area, the intensity of adoption of the wheat technology packages by adopters is computed following [88] as follows:

$${\text{AI}}_{\text{i}}={\sum }_{\text{i}=1}^{\text{n}}\frac{\frac{{\text{AIT}}_{\text{i}}}{{\text{RIT}}_{\text{i}}}}{\text{NP}}$$
(5)

where \({\text{AI}}_{\text{i}}\)= is the adoption index of the ith household; i = 1, 2, 3,…; \({\text{AIT}}_{\text{i}}\)= actually applied quantity of inputs; \({\text{RIT}}_{\text{i}}\)= is the recommended quantity of input; \(\text{NP}\)= number of practices; \(n\)= is the total number of household heads.

Depending on Eq. (5) above, the adoption index is fixed at below, 1%, above, or 100%, and all the elements of the packages are specified as follows:

$${\text{AI}}_{\text{i}}={\sum }_{\text{i}=1}^{\text{n}}\frac{\frac{{\text{AW}}_{\text{i}}}{{\text{AT}}_{\text{i}}}+\frac{{\text{FA}}_{\text{i}}}{{\text{FR}}_{\text{i}}} +\frac{{\text{ARP}}_{\text{i}}}{{\text{RRP}}_{\text{i}}} }{\text{NP}}$$
(6)

where i = 1, 2, 3… n; n = total number of households; AIi = adoption index of the ith household; AW = area under improved variety of wheat in the ith household; AT = total area covered wheat production (all seed types if any); FA = agrochemicals such as fertilizers applied per hectare; FR = agrochemicals fertilizer recommended per hectare; ARP = area under wheat row planting in hectare; RRP = total area of wheat planted both through row planting and broadcasting; and NP = number of practices.

3.5 Definitions of variables and hypotheses

3.5.1 Dependent variable

Adoption Index: It represents the adoption of wheat technology packages used in the model by computing the index of use and use intensity of technology packages. The packages include improved wheat varieties, wheat row planting, agrochemicals such as fertilizers (NPS and UREA), as specified by Eq. (6).

3.5.2 Independent variables

Independent variables selected from demographic, socioeconomic, institutional, and resource ownership factors were defined, hypothesized, and are represented in Table 2.

Table 2 Variables, descriptions and hypotheses
Full size table

4 Results and discussion

4.1 Descriptive results

The average ages of the sampled respondents in all districts were 40.1 with standard deviation of 8.44. This implies that, on average, wheat producer farmers’ ages were sufficient to be experienced in the adoption of various agricultural technology packages. When combined, all districts’ sampled households attended 3.798 schooling years on average. Households selected as a sample from all districts travel 32.79 and 14.24 min on average to reach the nearest input market and farm areas from their residential areas, respectively. Similarly, the average number of family members of sampled households in all districts was 4.46 persons on average.

The size of farm households has a direct or indirect effect on farmers’ adoption of agricultural technologies. Greater landholding size could be an input for improved all-over agricultural production if wisely used, and could also be a reason for the inefficiency of production. In the study area for all districts in combination, the average size of farm areas was 1.977 ha, with a standard deviation of 0.853. Among the farm households in the study area, 0.653 hectares of land were allocated for wheat cultivation. The number of households sampled for livestock in all districts was 6.21 on average, with a standard deviation of 1.77. Livestock is the source of many wheat production activities in areas where production machinery is rare. The average income which sampled households obtained from annual farm activities (agricultural production and livestock) for all districts were 52,875 Ethiopian Birr (ETB). The overall adoption index, which was used as a dependent variable in this study, is derived from different components of wheat production technology packages such as improved wheat varieties, agro-chemicals such as fertilizers (NPS, UREA and chemicals), and wheat row planting. The results show that the overall adoption index is 0.575 and the adoption index across each district does not vary significantly. The average adoption intensity of smallholder farmers in the study area is found to be low, caused by incomplete adoption of each recommended level of wheat technology packages. The low adoption rate of wheat technology packages could be attributed to factors related to the high cost of wheat production inputs and a lack of awareness about the technology packages, which caused a low improvement in wheat production (Table 3).

Table 3 Summary statistics of continuous variables used in the analysis.
Full size table

With regard to dummy variables, both male and female households participated in the production and adoption of wheat production technologies, although the number of females participating was low when compared with their male counterparts. On an average, 87.39% of the sampled wheat producers were male household heads. The proportion of male household heads that produce and adopt wheat production technology packages is high in Abe Dongoro district district (90.14%) when compared with other districts which was (89.16% and 85.16%) for Horo and Ababo Guduru districts, respectively). This difference among all districts of different agro-ecologies was not statistically significant (χ2 = 1.65), implying that no wheat production difference across each district was caused by differences in gender differences (Table 4).

Table 4 Summary statistics of dummy explanatory variables used in the analysis (%)
Full size table

On average, for all districts in the study area, 54.95% of the sampled households accessed credit services from different financial institutions located in each district’s main town, while the difference in access to credit services across the district was statistically significant at the 5% level, with χ2 = 6.73. Comparing each district, access to extension services was statistically significant at the 1% level with a value of χ2 = 9.37, implying that there is a difference in access to extension services in each district. All over the districts, about 41.82% of the sampled households had access to extension services on average, which indicates that for each district, differences in access to extension services occurred. Among the sampled households, lower proportions were members of different cooperatives, which might support them in producing and adopting wheat production technologies. Among the sampled households, only 49.59% were members of agricultural cooperatives. The difference was not statistically significant (χ2 = 0.54), showing that household location in different agro-ecologies (districts) had no effect on membership in agricultural cooperatives.

Differences in households’ location at different agro-ecologies with respect to access to improved wheat seed were statistically significant at the 5% significance level with the value of χ2 = 7.17, showing that the difference in district influences access to improved seeds. Overall, only 21.44% of farmers were accessed and purchased, requiring improved wheat seeds in the study area. The differences in participating in different agricultural training given in each district were also not statistically significant (χ2 = 0.26), implying that the difference in the location of households did not cause a difference in the training given. Among the sampled households, 46.11% of them had received training to improve agricultural production capacity. Among total sampled households, 50.67% had a good perception of their subsequent year’s wheat production, which enabled them to adopt wheat production technology packages, while the difference across each district was not statistically significant (χ2 = 4.27). About 45.84% of households engaged in off/non-farm income-generating activities in addition to agricultural production, while the difference in household location did not show a significant influence (χ2 = 0.90) on household engagement in off/non-farm activities (Table 4).

4.2 Econometric results

Table 5 presents the empirical findings of the two-limit Tobit econometric model. Prior to running a two-limit Tobit model, independent variables were tested for multicollinearity using Variance Inflation Factors (VIFs), and no multicollinearity problem was observed in these variables. The mean VIF was 1.29 with 1.04 and 2.48 VIF minimum and maximum, respectively (Appendix Table 6). The omitted variable (OV) test given in Appendix Table 7 shows that there was no problem with explanatory variable omission, as its P value is insignificant (0.7664). Similarly, the data were tested for the existence of heteroscedasticity, and the model has no heteroscedasticity problem as the significance value from the Breusch–Pagan/Cook-Weisberg test was 0.1619, which is insignificant (Appendix Table 7). The F-statistics showed that the model was a good fit at the 1% significance level. The index was computed from use intensity of technology packages such as improved seed use intensity, row planting use, and fertilizer (NPS and UREA). Seventeen explanatory variables were hypothesized to influence the adoption of wheat production technology packages represented in the index. Among the independent variables included in a two-limit model to affect the adoption intensity, educational level of household head, households’ location from the nearest market, access to credit services, access and utilization of improved seed, number of livestock owned, household production perception about wheat yield, household participation on farm training, and income from farm production significantly affected the adoption of wheat production technology packages at different significance levels.

Table 5 Adoption of wheat technology packages (Application of two-limit Tobit model).
Full size table

Unlike expected, differences in households’ locations due to different agro-ecologies had no significant influence on farmers’ adoption intensity of wheat technology packages which implies that households’ location did not contributed for differences in technology adoption rate. Farmers’ distance from their farm areas also had no effect on adoption intensity of wheat technology packages. Theoretically, services provided for farmers by extension workers are assumed to positively influence the adoption intensity of wheat technology package adoption, but unlike expected, access to extension services was not significant and its influential coefficient was negative. All other insignificant variables like land holding size, age, sex, household size, cooperative membership, and access to off/non-farm income were also assumed to affect wheat technology package adoption intensity, but had no significant effect on adoption intensity of wheat production technology packages.

The educational level of households significantly affected the adoption of technology packages at the 5% significance level, as expected. As the educational level of households increased by one year, the likelihood of wheat production package adoption increased by 1.52%, keeping other factors constant. This implies that educated farmers can easily be aware of the importance of technology adoption to adopt new technologies, easily get information about technologies and utilize them, be technically capable of getting training from agricultural experts, and apply the offered training on the ground. Overall, education provides the basis for acquiring, processing, sharing, using, and disseminating knowledge and information about wheat production technologies. In line with these findings, [28, 65, 82, 89] revealed that an increased level of households’ education increased adoption of improved maize variety, rice technologies, garden coffee production technology packages, and wheat production technology packages, respectively.

The econometric model results showed that households’ distance to the nearest market significantly and negatively influenced the adoption of wheat production technology packages at the 1% significance level. The results show that, as households’ walking to the nearest market increases by one minute, the likelihood of adopting wheat production technology decreases by 0.49%, keeping other factors constant. As market distance increases, farmers’ probability of adoption and the extent of adoption of wheat technology packages decrease. This is because market centers enable farmers to access agricultural production inputs and supply their produce to the market; however, if they are found far from the market, they become constrained to obtain these services. Similarly, as they are far from the market, they spend more time and incur greater costs in obtaining services that prevent them from adopting wheat production technology packages. This result is consistent with the findings of ref. [82], who showed that market distance and the probability of adoption and intensity of adoption of wheat production technology are negatively related.

As expected, households’ access to credit services from different sources positively affected the wheat technology package adoption intensity at the 10% significance level. The results of this study reveal that the adoption intensity of wheat technology packages by households that accessed credit increased by 5.21% compared to those that did not access. Access to credit services can solve the financial problems associated with purchasing wheat production inputs. Moreover, it enables households to search for recommended production technologies, capable of adopting them, and purchasing required inputs on time by minimizing constraints of input availability. This result is in line with the findings of ref. [65] who showed a positive effect of access to credit on households’ adoption of upland rice technologies.

Access to and purchase of improved wheat seed positively and significantly affected households’ adoption intensity of wheat production technology packages at a 1% significance level. Access to and purchase of improved wheat seed increases the adoption intensity of wheat production technology packages by 10.64%, keeping other factors constant. Accordingly, access to and purchase of available improved wheat seeds at the right time with the required quantity enables farmers to adopt different wheat production packages. The lack of access to inputs and imperfections in seed markets can limit the adoption of agricultural technologies. The findings of this study are in line with result of ref. [66], who revealed that supplying improved seed at the right time with the required quantity increases the farmer’s probability of adopting row planting for wheat production.

The number of livestock owned significantly and positively affected the adoption of wheat production technology packages at the 1% significance level. The results revealed that an increase in the number of livestock increased the adoption intensity of wheat production technology packages by 2.65%, keeping other factors constant. The results revealed that farmers who have a larger number of livestock can easily produce agricultural products and use them as a source of income to purchase agricultural inputs through their sales. This result is in line with the research findings of ref. [90], who discussed that the number of livestock households owned positively affects the adoption of improved cultivars.

Households’ perception of the productivity of wheat production positively and significantly influenced the adoption of production wheat technology packages at the 5% significance level. As households have a positive perception of wheat yield, the likelihood of adopting wheat production technology packages increased by 6.55%, keeping other factors constant. It is revealed that farmers who have a better perception of wheat yield than other crops will more likely adopt wheat-improved technology packages than those who do not assume that they will be beneficiary of the technology. They compare their perception of the yield of different crops they produce and are willing to adopt technologies for crops with higher yields than other crop yields. The findings of ref. [65] were consistent with the results of this study.

As expected, households’ participation in farm training had a positive effect on the adoption of wheat production technology. The coefficient of participation in farm training is statistically significant at the 5% level of significance, and it increases adoption intensity by 7.35%. The results suggest that training helps farmers to be aware of innovative and new ideas for input usage and improves their knowledge and insight for productivity maximization and technology adoption. This result is consistent with the findings of refs. [69, 91] who revealed that agricultural training had a positive effect on the adoption of teff and groundnut-improved seed technologies, respectively.

Similar to the hypothesis, annual farm income positively and significantly affected the adoption of wheat production technology packages at the 5% significance level. The results showed that, as the annual farm income of households increased by one birr wheat production technology, adoption increased by 7.36%, keeping other factors constant. This implies that households with higher annual farm income are more likely to adopt wheat technology packages than households with lower annual farm income. Moreover, it is the fact that income obtained from agricultural production enables farmers to improve their capacity to purchase agricultural production inputs and to be willing to adopt agricultural technologies easily. This result is in line with the research findings of refs. [71, 82], who revealed that annual farm income positively affected the adoption and intensity of adoption of improved wheat technologies.

5 Conclusions and recommendations

The adoption of agricultural technology packages is a policy concern for increasing and sustaining agricultural productivity to address agricultural production constraints and low productivity in Ethiopia. Technology packages adoption in wheat production resulted in the desired benefits to households when these packages were adopted in recommended combinations. This study analyzed the adoption of wheat production technology packages by using a two-limit Tobit model. The study used survey data were collected from a total sample of 373 farm households. The wheat production technology packages considered in this study include improved wheat variety, wheat row planting, and application of recommended agro-chemicals while a two-limit Tobit model was employed to analyze the adoption of these wheat technology packages. Moreover, several hypothesized demographic, institutional, farm, and resource ownership variables had different and significant effects on households’ adoption of technology packages. Accordingly, the educational level of the household head, distance from the nearest market, access to credit services, improved seed purchased and used, number of livestock owned, farmers’ perception about wheat yield, participation in farm training, and annual farm income are influential factors affecting the adoption of wheat technology packages, with different effects at different significance levels.

The study identified that scaling-up the adoption of wheat technology packages requires the adoption of all recommended wheat technology packages in combination rather than adopting a single or few technology packages to achieve the desired outcome of technology adoption. This suggests that farmers may face the problem of the high cost of adopting the packages, such as the cost of fertilizer (NPS and UREA), and they need practical advice on how to apply wheat row planting, the use of improved seeds, and the application of other technology packages. To achieve this, organizing and strengthening institutions that provide credit services and regular training for households to solve their financial problems for purchasing production inputs and improve their understanding of technology adoption are necessary. Therefore, any stakeholder who directly or indirectly engages in farmers’ institutional issues should provide advice to promote the adoption of wheat production technology packages. Additionally, farmers’ participation in farm training profoundly improves their awareness and willingness to adopt technology packages. Hence, government and non-government bodies should prepare regular training for farmers to improve their technology adoption skills and knowledge. The availability of wheat improved variety and households’ ability to purchase wheat are another challenges in enhancing the adoption of wheat technology packages, and the concerned institutions should solve the accessibility of the input and provide it to farmers at affordable prices.

Therefore, adoption intensity of wheat technology packages in the study area is low and its improvement needs continuous involvement of agricultural policymakers, qualified agricultural practitioners and development agents’ concerns in solving the identified wheat technology adoption barriers and subsiding farmers to solve their financial challenges through credit services. They should also devise a new and adaptable agricultural strategies to enable farmers adopt all recommended technology packages. The adoption of recommended agricultural technology package of agriculture needs further future researches which could result in improvement of agricultural productivity and overall wellbeing of smallholder farmers whose livelihoods are dependent on agriculture in Ethiopia.