Adoption analysis of agricultural technologies in the semiarid northern Ethiopia: A panel data analysis

Gebru, Menasbo; Holden, Stein Terje; Alfnes, Frode Article Adoption analysis of agricultural technologies in the semiarid northern Ethiopia: A panel data analysis Agricultural and Food Economics Provided in Cooperation with: Italian Society of Agricultural Economics (SIDEA) Suggested Citation: Gebru, Menasbo; Holden, Stein Terje; Alfnes, Frode (2021) : Adoption analysis of agricultural technologies in the semiarid northern Ethiopia: A panel data analysis, Agricultural and Food Economics, ISSN 2193-7532, Springer, Heidelberg, Vol. 9, Iss. 1, pp. 1-16, https://doi.org/10.1186/s40100-021-00184-6 This Version is available at: https://hdl.handle.net/10419/240295 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ RESEARCH Open Access Adoption analysis of agricultural technologies in the semiarid northern Ethiopia: a panel data analysis Menasbo Gebru 1,2* , Stein T. Holden 1 and Frode Alfnes 1 * Correspondence: [email protected] 1 School of Economics and Business, Norwegian University of Life Sciences, P. O. Box 5003, 1432 Ås, Norway 2 Department of Economics, Mekelle University, P. O. Box 451, Mekelle, Ethiopia Abstract Agricultural technology change is required in developing countries to increase the robustness to climate-related variability, feed a growing population, and create opportunities for market-oriented production. This study investigates technological change in the form of adoption of improved wheat, drought-tolerant teff, and cash crops in the semiarid Tigray region in northern Ethiopia. We analyze three rounds of panel data collected from smallholder farms in 2005/2006, 2009/2010, and 2014/2015 with a total sample of 1269 households. Double-hurdle models are used to assess how the likelihood (first hurdle) and intensity of technology adoption (second hurdle) are affected by demographic, weather, and market factors. The results indicate that few smallholders have adopted the new crops; those that have adopted the crops only plant small shares of their land with the new crops, and that there has been only a small increase in adoption over the 10-year period. Furthermore, we found that high population density is positively associated with the adoption of improved wheat, and previous period’s rainfall is positively associated with the adoption of drought-tolerant teff. The adoption of cash crops is positively associated with landholding size and access to irrigation. The policy implications of these results are that the government should increase the improved wheat diffusion efforts in less dense population areas, make sure that drought-tolerant teff seed is available and affordable after droughts, and promote irrigation infrastructure for production of cash crops. Keywords: Semiarid areas, Climate risk, New crop varieties, Double-hurdle, Northern Ethiopia JEL Classification: O33, Q12, Q16, R34 Background Adoption of improved agricultural technologies is an important means of adapting to climate change, improving agricultural productivity, and facilitate the transition from subsistence agriculture to market-oriented agriculture (Bezu et al. 2014; De Janvry and Sadoulet 2002; Mendola 2007; Minten and Barrett 2005; Yu et al. 2011; Zilberman et al. 2012). Among the technologies adopted by farmers in the Ethiopian highlands are improved wheat, drought-tolerant teff,and cash crops (Belay et al. 2006; Shiferaw © The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Agricultural and Food Economics Gebru et al. Agricultural and Food Economics (2021) 9:12 https://doi.org/10.1186/s40100-021-00184-6 et al. 2014; Wale and Chianu 2015). In this paper, we investigate to what extent farmers in the semiarid Tigray region of Ethiopia have adopted improved wheat, droughttolerant teff,and cash crops and which factors explains the adoption and intensity of adoption. Technology diffusion often takes years and can best be captured using panel data. However, most studies on the adoption of improved wheat in semiarid agriculture in Ethiopia use cross-sectional data (Kelemu 2017;Kotuetal.2000; Lobell et al. 2005; Matuschke et al. 2007;Shiferawetal.2014; Tesfaye et al. 2016). One of the few studies including a time dimension is Abera’s(2008), which used cross-section household data from 2001 with recall data back to 1997 and estimated factors affecting adoption of improved wheat in northern and west Shewa zones of Ethiopia. He analyzed how farmer and farm characteristics are correlated with adoption and intensity of adoption, but does not cover important supply-side constraints that need attention. Studies of drought-tolerant teff in Ethiopia include Wale and Chianu (2015) and Belay et al. (2006). Wale and Chianu (2015)examinedfarmers’demand for drought-tolerant teff using cross-sectional data. The study of Belay et al. (2006) used data from an experiment on village demonstration plots, including 41 farmers in 2002 and 2003, and found that farmers adopt drought-tolerant teff varieties when there is limited rainfall. To the best of our knowledge, empirical studies of the adoption of drought-tolerant teff using rich panel data from semiarid agriculture are missing. Adoption of cash crops is mainly associated with access to irrigation and has a dual advantage. First, irrigation and adoption of cash crops typically allow the smallholders to harvest more than one time per year, which lead to improved land productivity. Second, the adoption of cash crops leads to improved output market integration and increased income. Ethiopia has adopted smallholders’commercialization as part of its economic transformation strategy (Gebremedhin et al. 2009). The development of irrigation reduces the production risk in semiarid areas and expansion of public investments in infrastructures improve market access. This has improved agricultural productivity and enhanced market participation by Ethiopian smallholders (Gebregziabher et al. 2009; Hailua et al. 2015). The main contribution of this study is threefold: first, we provide new insight into the development in the adoption of the three improved agricultural technologies improved wheat, drought-tolerant teff, and cash crops in Tigray, Ethiopia. Second, we provide new insight into factors affecting the likelihood of adoption and intensity of adoption for these improved agricultural technologies. Third, we discuss policy implications for how to best integrate and reap the benefits from the promotion of improved wheat, drought-tolerant teff and cash crops, given their importance for food productivity, food security, and market integration. Theoretical framework Household’s adoption decision of new technology is usually modeled as a choice between traditional and new technology. A farm household adopts the new agricultural technology when the expected benefit from adoption is higher than Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 2 of 16 without adoption (Amare et al. 2012;Bezuetal.2014;MaandShi2015). More recently, the literature has started to investigate constraints that could cause only partial adoption across and within farms. The theoretical framework of this study builds on the state-contingent partial adoption framework for new technologies in a risk-exposed economy, as in Holden and Quiggin (2017). Partial and state-contingent adoption reflects that household choices may be affected by factors such as stochastic weather events, market imperfections in input and output markets, limited knowledge about the performance of new technologies under different states of nature, limited availability and high cost of technologies, and heterogeneity in farm and household characteristics. Climate change and climate risk may affect technology adoption as illustrated by the state-contingent production approach (Holden and Quiggin 2017). This Table 1 Summary of statistics of variables used in the analysis by survey year (mean values) Variables description 2006 2010 2015 Pooled Mean St. Err. Mean St. Err. Mean St. Err. Mean St. Err. Three subsamples High-yield wheat adoption (yes =1) 0.129 0.018 0.184 0.018 0.138 0.016 0.151 0.010 High-yield wheat area planted, adopters (tsimidi) 0.158 0.029 0.306 0.041 0.199 0.031 0.225 0.020 Drought-tolerant teff adoption (yes=1) 0.060 0.013 0.039 0.009 0.160 0.017 0.091 0.008 Drought-tolerant teff area planted, adopters (tsimidi) 0.097 0.026 0.087 0.027 0.233 0.040 0.145 0.019 Cash crop adoption (yes =1) 0.115 0.017 0.184 0.018 0.160 0.017 0.156 0.010 Cash crop area planted, adopters (tsimidi) 0.044 0.008 0.190 0.034 0.191 0.037 0.150 0.019 Owned land (tsimidi) 4.430 3.261 4.429 3.093 4.542 2.928 4.472 3.077 Full sample Farm-level population pressure 2.168 0.176 1.778 0.075 2.100 0 .330 2.007 0.136 Mean value of farm-level pop. pressure at community level 2.091 0.076 1.987 0.061 1.965 0.058 2.007 0 .037 Distance to market (h) 1.407 0.048 1.401 0.043 1.394 0.043 1.400 0.025 Mean rainfall of 12 years panel (cm) 47.057 0.953 45.203 0.710 44.948 0.673 45.615 0.441 Rainfall variability (Std. Dev.) of 12-year panel (cm) 8.566 0.069 8.820 0.054 8.838 0.052 8.757 0.033 One-year lagged positive deviation from long-term mean rainfall (cm) 0.000 - 1.287 0.110 15.890 0.650 6.457 0.323 One-year lagged negative deviation from long-term mean rainfall (cm) 10.612 0.175 5.199 0.199 0.000 - 4.717 0.145 Two-year lagged positive deviation from long-term mean rainfall (cm) 0.000 - 0.991 0.119 2.817 0.133 1.410 0.072 Two-year lagged negative deviation from long-term mean rainfall (cm) 14.504 0.363 7.011 0.305 1.335 0.104 6.919 0.211 Sample size Improved wheat 187 287 340 814 Drought-tolerant teff 218 336 441 995 Cash crops 31 126 141 298 Source: NMBU and MU household panel Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 3 of 16 approach states that farmers’adoption decision depends on their perception of risk associated with the choice of the new technology relative to alternative technologies and the states of nature that may be realized after adoption decisions are made. Limited knowledge of the performance of new technologies under alternative states of nature may be one constraint. Partial adoption and exposure to different states of nature can over time help farmers build realistic and more accurate expectations about alternative technologies and thereby influence the adoption and adaptation process. Hence, households exposed to earlier weather shocks and who are risk-averse are more likely to choose a less risky technology such as drought-tolerant crop varieties when they have developed their knowledge about these (Amare et al. 2012;Antle1987; Holden and Quiggin 2017). Another research string important for our study is the literature on technology diffusion. Pan et al. (2018) investigated how technology diffusion processes affect farmers’adoption decisions. They found that factors making it easy to learn about the benefits of new technologies have a positive impact on adoption rates. Examples of such factors are extension services, field demonstrations, market integration, and viewing and learning from other farmers. Other studies also point to learning externalities, social learning diffusion, communication patterns, and following successful neighbors’practices as drivers of technology diffusion (Conley and Udry 2010;Geniusetal.2014). In total, these studies point in the direction of a gradual increase in adoption of improved agricultural technologies over time, if they are available and affordable. Based on the theoretical framework, we propose the following hypotheses for testing: H1: There is a gradual increase in the adoption and intensity of adoption of the three improved agricultural technologies over the 10-year time period. H2: Improved wheat is more likely to be adopted in areas with high population pressure and by more land-constrained households (high farm-level population pressure). H3: Drought-tolerant teff is more likely to be adopted in areas with more rainfall variability and in areas exposed to recent rainfall shocks (droughts). H4: Cash crops are more likely to be adopted in areas with good market access (short distance to markets). Method Survey design and data The data are collected in Tigray in northern Ethiopia. The region is semiarid and exhibits high population pressure (Appendix Table 5), seasonal and erratic rainfall, relatively low agricultural potential, and limited access to sizeable markets. The data used in this study come from three rounds of farm household surveys conducted in 2005/ 2006, 2009/2010, and 2014/2015 production seasons (Table 1). The panel sample is based on a survey conducted in 1998/1999 using a twostage sampling technique and described in Hagos and Holden (2003). In the first 1 These are available online IRI/LDEO Climate Data Library: http://iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NCEP/.CPC/.FEWS/.Africa/.DAILY/.ARC2/.. Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 4 of 16 stage, communities were selected from the rural districts of the region to reflect differences in agricultural potential, population density, agroecology, market access, and access to irrigation. In the second stage, 25 households were randomly sampled from a list of farm families in the selected communities for detailed interviews. Most of the technologies of interest for this study were introduced in the study region after year 2000, and we use data from the three survey rounds in 2006, 2010, and 2015, each covering the previous year’s cropping seasons. Over time, some households dropped out of the sample, and new were added, resulting in an unbalanced household panel. To examine farmers’technology adoption decisions, we use information on household and farm characteristics including land and non-land endowments, farm-level population pressure, indicators of access to infrastructure (marketplace and road), and rainfall at community level. We construct long-term average annual rainfall, variation (standard deviation) in average annual rainfall, and 1- and 2-year lagged annual rainfall at the community level from the monthly satellite record of the African Rainfall Climatology Version 2 (ARC2) for the years 2003– 2014 1 . Presuming that access to technology differs according to the features of agroecology and accessibility of public services, we divide the households into three access-to-agricultural-technologies groups. The first access group is households residing in the mid and highland agroecology with access to improved wheat (Group 1). In Ethiopia, wheat is a mid and highland crop (Doss et al. 2003;Kotu et al. 2000) and is distributed to households in this agroecology. The second access group is households who live in drought-affected agroecologies with access to drought-tolerant teff (Group 2). Promotion of the adoption of drought-tolerant teff is an important strategy for adapting to the changing climate in these areas. The third access group is households who live in communities with access to irrigation and, thereby, are able to grow cash crops (Group 3). Access to irrigation such as a dam or groundwater that can be used to grow crops facilitate the adoption of cash crops. We will later refer to these three regionally determined access groups as the households with access to improved wheat, access to droughttolerant teff,and access to cash crops, respectively. Estimation method: double-hurdle model The technology adoption literature proposes various econometric methods that can be used in modeling the behavior of households’demand for new agricultural technology and identify the factors that can explain adoption decisions (Heckman 1979; Maddala and Nelson 1975; Wooldridge 2010). We present results based on Cragg’s double-hurdle models that allow variables to have different effects on adoption and intensity of adoption. In the first hurdle, we estimate a probit model to determine the probability that the households adopt the new agricultural technologies. In the second hurdle, we use a truncated regression model to determine the intensity of the adoption. We estimate the double-hurdle models for the adoption of the three technologies separately using the subsample that has access to the respective technologies. Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 5 of 16 We first run parsimonious models with the key explanatory variables of interest: household level and average community-level population pressure (family size/farm size), average community level rainfall and rainfall variability over the last 12 years, 1 and 2 years of lagged deviations from average rainfall, distance to market, and farm-level access to irrigationinthecaseofcashcrops.Wethenassess the robustness of these results by including additional household control variables with and without a correlated random effects (CRE) approach (see elaboration below). The control variables include household head characteristics (gender, age, age squared, and literacy status), family labor (number of adult males and females), household resource endowments (number of oxen, mobile phone ownership (dummy), and size of owned land). Two-year dummies are also included to capture change over time (2010 and 2015). We will refer to these control variables by the vector X. We specify the following Craggit double-hurdle model: Hurdle 1: Probability of adoption, binary probit Pw ijt ¼1  ¼αpPijt þαrRct þαdDct þαnXijt þγnXij  þuiþeijt ð1Þ Hurdle 2: Intensity of adoption, truncated regression model Yijt ¼βpPijt þβrRct þβdDct þβnXijt þδnXij  þμiþεijt if w ¼1;0 otherwise;ð2Þ where w ijt is a variable indicating whether or not the household adopt the new technology, taking the value of 1 if the household adopts the technology and 0 otherwise; Y ijt is the observed intensity of adoption measured as the log of area planted with the technology for the households that have adopted the technology; P ijt represents household and community population pressure; R ct is a vector representing the rainfall variables; D ct is the distance to market; and X ijt is a vector of the control variables as explained above. To control for unobserved heterogeneity, the means of the time-varying Xvariables, Xij;are included, which is the Mundlak (1978) and Chamberlain (1982), approach, also known as the correlated random effects (CRE) approach (Wooldridge 2010). This approach controls for other time-constant unobservable variables in a similar way as household fixed effects do in a linear panel data model. i,j,andtare individual household, technology type, and time identifiers, respectively; αand βare the parameters to be estimated for the nX-variables, and u i and μ i are normally distributed random effects, constant for each household over time; e ijt and ε ijt are error terms assumed to be independent and normally distributed, e ijt ~N(0, 1) and ε ijt ~N(0, σ 2 ). A limitation of the CRE approach is that it takes many degrees of freedom and that may affect significance levels in small samples such as in the second stage of our double-hurdle models. We, therefore, run models without and with this specification as a robustness check. We have also tested for attrition bias, but found no significant effect on our results, and hence report the results without attrition controls. Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 6 of 16 Results Descriptive analysis Table 1presents the mean values of technology adoption rates and intensity of adoption by technology and year in our panel, as well as the key variables of interest for our study. The adoption rates measure the share of households using each crop in the region they are available, while the adoption intensity measures the area the adopters planted with each crop. The areas are measured in tsimdi;one tsimdi is approximately 0.25 ha. Average farm size in tsimdi is also included in the table, for comparison with areas planted with the new crops of interest to our study. We observe that the adoption rate for the improved wheat increased from 12.9% in 2006 to 18.4% in 2010 and decreased to 13.8% in 2015, indicating an initial increase and then stagnation and decline in adoption. The pattern for adoption intensity shows a similar trend over time. On average across years, adopters of improved wheat had planted about 5% of their farm area with improved wheat. Drought-tolerant teff had adoption rates of 6, 3.9, and 16%, respectively over the 3 years, indicating a stagnant low rate first but then a substantial increase in the adoption rate. The adoption intensity was stagnant and small from 2006 to 2010 but then more than doubled from 2010 to 2015. On average across years, adopters of drought-tolerant teff had planted about 3% of their farm area with droughttolerant teff. For cash crops, we see an initial increase in adoption rate from 11.5 to 18.4%, and then a weak decline to 16%. On average across years, adopters of cash crops had planted about 3% of their farm area with cash crops. Overall, we see low adoption rates and only small shares of the farms of adopters covered by the new crops. Only for drought-tolerant teff do we see a clear trend towards increasing adoption. For the two other technologies we see a stagnation or decline in the adoption rates over time. Hence, we do not find support for our Hypothesis H1 stating, “There is a gradual increase in the adoption and intensity of adoption of the three improved agricultural technologies over the 10-year time period.” Estimation results The results of the double-hurdle model for adoption and intensity of adoption are presented in Table 2for improved wheat, Table 3for drought-tolerant teff, and Table 4for cash crops. We discuss one technology at a time in the following three sections. The three technologies are largely adopted in different areas and do, to a very small extent, compete for the same land. We can, therefore, consider their adoption as independent processes. The adoption for each technology is estimated for the areas that have access to these technologies and where these technologies are suitable. To verify whether the results are robust, we present the results from three different double-hurdle models for each technology. The first is a parsimonious version that includes only the key variables of interest, the second includes additional controls, and the third includes the means of the RHS variables Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 7 of 16 including additional controls (CRE approach). In our interpretation, we give most weight to the results that are significant across all three model versions. We focus primarily of the assessment of our four hypotheses in the interpretation of the results. Improved wheat adoption The results for the improved wheat models are presented in Table 2.OurHypothesis H2 stated, “Improved wheat is more likely to be adopted in areas with high population pressure and by more land-constrained households (high farmlevel population pressure)”.Table2shows that farm-level population pressure is strongly and robustly positively correlated with adoption of improved wheat. This result is significant at 1% level in two of three model variants, and significant at Table 2 Double-hurdle estimation factors affecting adoption of improved wheat (Craggit model) Variables Without HH controls With HH controls HH controls + CRE Hurdle 1 Hurdle 2 Hurdle 1 Hurdle 2 Hurdle 1 Hurdle 2 Mean farm level pop. pressure at community level 0.018 (0.040) −0.163*** (0.038) 0.052 (0.046) −0.069** (0.034) 0.049 (0.052) −0.054 (80.035 Deviation of farm level pop. pressure from community mean 0.013** (0.005) -0.003 (0.003) 0.014*** (0.005) 0.000 (0.001) 0.017*** (0.006) 0.000 (0.001) Mean rainfall 2003–2014 (cm) −0.042** (0.017) −0.010 (0.017) −0.052*** (0.018) −0.006 (0.014) −0.059*** (0.018) −0.008 (0.015) St. Dev. rainfall 2003–2014 (cm) 0.179** (0.100) 0.011 (0.100) 0.235** (0.107) 0.055 (0.090) 0.279*** (0.108) 0.063 (0.091) One-year lagged positive deviation rainfall (cm) 0.011 (0.012) 0.008 (0.010) 0.013 (0.012) 0.005 (0.009) 0.017 (0.012) 0.007 (0.010) One-year lagged negative deviation rainfall cm) −0.018 (0.018) 0.004 (0.014) −0.021 (0.018) −0.008 (0.014) −0.022 (0.018) −0.004 (0.014) Two-year lagged positive deviation rainfall (cm) 0.001 (0.023) 0.030 (0.019) 0.005 (0.024) 0.017 (0.018) 0.006 (0.024) 0.016 (0.019) Two-year lagged negative deviation rainfall (cm) 0.030** (0.012) −0.003 (0.007) 0.033*** (0.012) 0.000 (0.006) 0.031** (0.012) 0.001 (0.007) Distance to market (h) −0.056 (0.073) 0.072 (0.044) -0.054 (0.075) 0.048 (0.038) −0.060 (0.075) 0.056 (0.037) Year 2010 dummy 0.174 (0.157) 0.219** (0.107) −0.016 (0.171) 0.168 (0.111) 0.072 (0.178) 0.227** (0.124) Year 2015 dummy −0.055 (0.259) 0.097 (0.170) −0.217 (0.262) 0.002 (0.157) −0.304 (0.263) 0.061 (0.165) Constant −0.755* (0.451) 1.105*** (0.324) −2.212** (0.928) −0.605 (0.555) −2.354** (1.136) −0.624 (0.653) Sigma constant 0.427*** (0.030) 0.380*** (0.024) 0.371*** (0.024) Chi 2 35.04 75.57 84.29 Log-likehood −514.01 −473.97 −461.78 Prob > Chi 2 0.0000 0.0000 0.0000 6.3.1.1.1. N814 192 814 192 814 192 Hurdle 1 is the probability of adoption, and hurdle 2 is the log of area planted upon adoption of the technology. The HH control + CRE models include the mean and actual value of time-variant household control variables (not reported in this table to save space). Numbers in parenthesis are cluster robust standard errors. ***, **, and * are to 1, 5, and 10% levels of significance, respectively. Source: NMBU and MU household panel survey Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 8 of 16 Amare M, Asfaw S, Shiferaw B (2012) Welfare impacts of maize–pigeonpea intensification in Tanzania. Agric Econ 43(1):27–43 https://doi.org/10.1111/j.1574-0862.2011.00563.x Antle JM (1987) Econometric estimation of producers’risk attitudes. Am J Agric Econ 69(3):509–522 https://doi.org/10.23 07/1241687 Bachewe F, Berhane G, Hirvonen K, Hoddinott J, Hoel J, Tadesse F, Yohannes Y (2014) Feed the Future (FtF) of Ethiopia– baseline report 2013. Ethiopia Strategy Support Program and International Food Pol Res Inse, Addis Ababa http://essp. ifpri.info/files/2014/12/FTF-baseline-report-FINAL-for-website.pdf Belay G, Tefera H, Tadesse B, Metaferia G, Jarra D, Tadesse T (2006) Participatory variety selection in the Ethiopian cereal tef (Eragrostis tef). Exp Agric 42(1) https://doi.org/10.1017/S0014479705003108 Bezu S, Kassie GT, Shiferaw B, Ricker-Gilbert J (2014) Impact of improved maize adoption on welfare of farm households in Malawi: a panel data analysis. World Dev 59:120–131 https://doi.org/10.1016/j.worlddev.2014.01.023 Cafer AM, Rikoon JS (2018) Adoption of new technologies by smallholder farmers: the contributions of extension, research institutes, cooperatives, and access to cash for improving tef production in Ethiopia. Agric Hum Values 35(3):685–699 Chamberlain G (1982) Multivariate regression models for panel data. J Econ 18(1):5–46 https://doi.org/10.3386/t0008 Conley TG, Udry CR (2010) Learning about a new technology: pineapple in Ghana. Am Econ Rev 100(1):35–69 https://doi. org/10.1257/aer.100.1.35 De Janvry A, Sadoulet E (2002) World poverty and the role of agricultural technology: direct and indirect effects. J Dev Stud 38(4):1–26 https://doi.org/10.1080/00220380412331322401 Dixon J, Nalley L, Kosina P, La Rovere R, Hellin J, Aquino P (2006) Adoption and economic impact of improved wheat varieties in the developing world. J Agric Sci 144(6):489–502 https://doi.org/10.1017/S0021859606006459 Doss CR, Mwangi W, Verkuijl H, De Groote H (2003) Adoption of maize and wheat technologies in Eastern Africa: a synthesis of the findings of 22 case studies: CIMMYT Ethiopian Agricultural Transformation Agency (2017). A roadmap towards the transformation of the Ethiopia seed. Seed system development strategy. Vision, systemic challenges, and prioritized interventions. Working strategy document. Available at http://extwprlegs1.fao.org/docs/pdf/eth172079.pdf. Accessed 24 Jan 2020. Firew M, Mizan A, Agidew B, Tafa J, Yeshitila M, Getachew T, Million E, Getachew A, Abdulesemed A, Daneil M (2016) Crop variety registry. Issu NO 19/2016. Minstry of AGrilcture and Natural Resource, Addis Abab Available at http://publication. eiar.gov.et:8080/xmlui/bitstream/handle/123456789/2404/CROP%20V. Accessed 2 Feb 2020 Gebregziabher G, Namara RE, Holden S (2009) Poverty reduction with irrigation investment: An empirical case study from Tigray, Ethiopia. Agric Wat Mang 96(12):1837–1843 https://doi.org/10.1016/j.agwat.2009.08.004 Gebremedhin B, Jaleta M, Hoekstra D (2009) Smallholders, institutional services, and commercial transformation in Ethiopia. Agric Econ 40(s1):773–787 https://doi.org/10.1111/j.1574-0862.2009.00414.x Genius M, Koundouri P, Nauges C, Tzouvelekas V (2014) Information transmission in irrigation technology adoption and diffusion: Social learning, extension services, and spatial effects. Am J Agric Econ 96(1):328–344 https://doi.org/10.1093/a jae/aat054 Hagos, F., & Holden, S. (2003). Rural household poverty dynamics in northern Ethiopia 1997-2000: analysis of determinants of poverty.Paper presented at the Chronic Poverty Research Centre International Conference, Manchester. Hailua G, Manjureb K, Aymutc K-M (2015) Crop commercialization and smallholder farmers livelihood in Tigray region, Ethiopia. J Dev Agric Econ 7(9):314–322 https://doi.org/10.5897/JDAE2019.1134 Heckman JJ (1979) Sample selection bias as a specification error. Econometrica J Econom Soc:153–161 Holden ST, Quiggin J (2017) Climate risk and state-contingent technology adoption: shocks, drought tolerance, and preferences. Eur Rev Agric Econ 44(2):285–308 https://doi.org/10.1093/erae/jbw016 Kelemu K (2017) Determinants of farmers access to information about improved wheat varieties: case of farmers in major wheat growing regions of Ethiopia. Int J Res Agric Sci 4(1) Kotu BH, Verkuijl H, Mwangi W, Tanner D (2000) Adoption of improved wheat technologies in Adaba and Dodola Woredas of the Bale Highlands, Ethiopia: CIMMYT Lobell DB, Ortiz-Monasterio JI, Asner GP, Matson PA, Naylor RL, Falcon WP (2005) Analysis of wheat yield and climatic trends in Mexico. Field Crop Res 94(2-3):250–256 https://doi.org/10.1016/j.fcr.2005.01.007 Ma X, Shi G (2015) A dynamic adoption model with Bayesian learning: an application to US soybean farmers. Agric Econ 46(1):25–38 https://doi.org/10.1111/agec.12124 Maddala, G. S., & Nelson, F. (1975). Switching regression models with exogenous and endogenous switching.Paper presented at the Proceedings of the American Statistical Association. Matuschke I, Mishra RR, Qaim M (2007) Adoption and impact of hybrid wheat in India. World Dev 35(8):1422–1435 https:// doi.org/10.1016/j.worlddev.2007.04.005 Mendola M (2007) Agricultural technology adoption and poverty reduction: a propensity-score matching analysis for rural Bangladesh. Food Policy 32(3):372–393 https://doi.org/10.1016/j.foodpol.2006.07.003 Minten, B., & Barrett, C. B. (2005) Agricultural technology, productivity, poverty and food security in Madagascar. Productivity, Poverty and Food Security in Madagascar (January 2005). https://ssrn.com/abstract=716142 or https://doi.org/10.2139/ ssrn.716142. Mundlak Y (1978) On the pooling of time series and cross section data. Econometrica J Econom Soc:69–85 Pan Y, Smith SC, Sulaiman M (2018) Agricultural extension and technology adoption for food security: evidence from Uganda. Am J Agric Econ 100(4):1012–1031 https://doi.org/10.1093/ajae/aay012 Shiferaw B, Kassie M, Jaleta M, Yirga C (2014) Adoption of improved wheat varieties and impacts on household food security in Ethiopia. Food Policy 44:272–284 https://doi.org/10.1016/j.foodpol.2013.09.012 Spielman DJ, Kelemwork D, Alemu D (2012) Seed, fertilizer, and agricultural extension in Ethiopia, Food and agriculture in Ethiopia: Progress and policy challenges, pp 84–122 Available at http://extwprlegs1.fao.org/docs/pdf/eth172079.pdf. Accessed 20 Jan 2020 Tesfaye S, Bedada B, Mesay Y (2016) Impact of improved wheat technology adoption on productivity and income in Ethiopia. Afr Crop Sci J 24(s1):127–135 Available at http://www.bioline.org.br/abstract?cs16022. Accessed 20 Jan 2019 Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 15 of 16 Wale E, Chianu J (2015) Farmers’demand for extra yield from improved tef [(Eragrostis tef (Zucc.) Trotter)] varieties in Ethiopia: implications for crop improvement and agricultural extension. J Agric Sci Technol 17(6) http://journals.moda res.ac.ir/article-23-5111-en.html Wooldridge, J. M. (2010) Econometric analysis of cross section and panel data: MIT press. Available at https://pdfs.sema nticscholar.org/ac93/a61c65fe3af707f19b7446f48756e4c7bd60.pdf. Accseed 12 June 2018. Wubeneh NG, Sanders J (2006) Farm-level adoption of sorghum technologies in Tigray, Ethiopia. Agric Syst 91(1):122–134 https://doi.org/10.1016/j.agsy.2006.02.002 Yu, B., Nin-Pratt, A., Funes, J., & Gemessa, S. A. (2011) Cereal production and technology adoption in Ethiopia. Ethiopia Strategy Support Program II (ESSP II), ESSP II Working Paper, 31. Zilberman D, Zhao J, Heiman A (2012) Adoption versus adaptation, with emphasis on climate change. Ann Rev Resour Econ 4(1):27–53 1941-1340/12/1010-0027$20.00 Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Gebru et al. Agricultural and Food Economics (2021) 9:12 Page 16 of 16