This study examines how farm households' decision to use modern agricultural inputs is influenced by the weather variability using a household panel survey merged with long‐term historical weather data from Ethiopia. As a part of an effort to fill methodological gaps that are observed and anticipated in previous similar literature, this study employs a more flexible modeling approach under a multiagricultural technologies framework. Findings suggest that weather uncertainty reduces the probability and intensity of adoption of productivity‐enhancing inputs, including chemical fertilizer and improved seed. The average partial effect for rainfall variability indicates that each additional percent of coefficient of variation of precipitation decreases the fertilizer use per hectare by, on average, 2.5%, other factors being constant. Similarly, a 1% increase in rainfall variability is associated with on average 0.6% and 4.5% decrease in probability and extent of improved seed use, respectively. We also observe that abundance rainfall during the previous production period increases the use of yield‐enhancing inputs in the current growing season. On the other hand, we observe that variability in rainfall increases the probability of the adoption of loss‐reducing agricultural technology, such as irrigation. Findings also indicate that by deterring their incentive to invest in productivity‐enhancing agricultural inputs, weather risk could confine uninsured farming households in a low productivity‐low income trap. Results also highlight the importance of policy interventions aiming to improve the risk‐bearing capacity of smallholder farmers in encouraging investment in productivity‐enhancing technologies.

Background Among others, the productive use of surplus labor is a viable mechanism to transform the agricultural sector and thus the whole economy in low-income countries. It is critically important to understand the factors that condition labor productivity to design and deploy effective agricultural and labor market policies. A few studies confirm that, at low-income levels, improving nutrition can contribute to the labor productivity of households. These studies rely heavily on self-reported farm data, which are prone to systematic and random measurement errors. The empirical evidence on this topic remains inadequate and inconclusive for this reason. Here, we substantiate whether better nutritional status enhances the labor productivity of farm households using objective measures of plot-level data from a recent household survey in Ethiopia. We also employ alternative measures of nutrition status indicators known as, Food Consumption Score (FCS) and Household Dietary Diversity Score (HDDS), inter alia, to capture additional aspects of nutrition such as diet quality and diversity, which are overlooked by calorie intake data. To deal with possible endogeneity, we employ a panel fixed effect estimation technique with a rich set of household socioeconomic and plot characteristics. Results We observe that the impact of current nutritional status, as measured by HDDS, on labor productivity varies considerably depending on the initial level of diet quality and diversity with a stronger and positive effect for low-consumption households. In an alternative specification, we also observe a positive farm labor productivity effect of current nutritional status as measured by FCS with a homogenous effect across households. However, the effect of the outcome of past nutritional status as evaluated by the Activity of Daily Living Index (ADLI) seems negligible. Conclusion Our findings indicate that improving nutrition can contribute to farm labor productivity at least for households with low current diet quality and diversity. Also, based on the findings, we conclude that there is a possibility of a low consumption–low productivity trap in Ethiopia.