![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
Adoption of chemical fertilizers is a high-risk and high-return investment option for smallholder agricultural households that heavily rely on rainfall. I document a persistent gap of above 10% in the adoption of chemical fertilizer between male- and female-headed smallholder farmers in Ethiopia. This gender gap remains after accounting for household characteristics, access to complimentary farm inputs, access to credit, soil quality, and crop selection. Using historical variability of rainfall at the district level as a measure of a district’s risk of crop failure, I find strong evidence that the gender gap in fertilizer adoption increases with the level of risk in the district. I explore the role of two competing hypotheses to explain this observation: gender difference in risk aversion and differential access to consumption smoothing/liquidity constraints by male- and female-headed households. I find strong evidence that gender differences in access to consumption smoothing/liquidity constraints play a minimal role, implying that gender difference in risk aversion plays the dominant role. This is consistent with a bulk of lab and field experimental studies that find evidence that women tend to be more risk-averse than men.
Acknowledgements
I am indebted to Sheetal Sekhri and James Harrigan for their helpful comments. I also thank Amalia Miller and Charles Holt for their comments.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 See Doss and Morris (Citation2001), for instance.
2 See Duflo, Kremer, and Robinson (Citation2008) for instance.
3 I also consider households’ choice between chemical fertilizer and no chemical fertilizer.
4 Based on Ethiopian Socioeconomic Survey (2011–2015), only 6% of the plots farmed were irrigated.
5 There are also studies that document that attitudes towards risk could change significantly, particularly when individuals are exposed to extreme shocks. See for instance Voors et al. (Citation2012) and Cassar, Healy, and von Kessler (Citation2017).
6 However, the result is robust to using alternative measures, such as the log of differences between (i) maximum and minimum, (ii) 90th and 10th percentiles, and (iii) 70th and 30th percentiles.
7 While the OLS includes more village level control variables, I limit the variable list to only the above list due to problem in convergence of the matching estimation (the village characteristics that are left out are highly correlated with the ones included).
8 See Korecha and Barnston (Citation2007) for a detailed discussion on predictability of rainfall and other climatic conditions in Ethiopia.
9 The survey asks if the household has access to credit service, and if the household doesn’t have access, the main reason for not having access.
10 I also use the number of oxen owned as an alternative. Results are quite similar.
11 However, there is significant labor and time investment required to produce manure and transport to the farm field.
12 The minor differences of point estimates from the corresponding tables are because I use zone fixed effects instead of district fixed effects to produce these figures. This was dictated because district fixed effects would subsume the marginal effects.
13 Related strands of literature suggest alternative mechanisms through which market exposure affect households’ risk preferences. For instance, List (Citation2003) argues that market exposure eliminates behavioural anomalies, such as endowment effects. Several related studies argue that market exposure improves rational decision making (Cecchi & Bulte, Citation2013; List & Millimet, Citation2008).
14 I am not attempting to identify the detailed mechanisms through which improved market access mute the role of risks, as doing so is difficult with observational data.