The Effects of Risk Preferences on the Adoption of Post-Harvest Technology: Evidence from Rural Cambodia

Abstract

We investigate how rice farmers’ risk preferences affect the adoption of post-harvest technology in Cambodia, focusing on moisture meters. We find that risk-averse farmers are more likely to adopt moisture meters, although the degree of farmers’ loss aversion or the extent to which they tended to overweight small probabilities do not affect the adoption. In the setting of our study, the effects of farmers’ risk preferences on agricultural technology adoption can be described by using expected utility theory. However, controlling for loss aversion and probability weighting improves the precision of examining the effects of farmers’ risk preferences on adoption.

Acknowledgements

We are deeply grateful to two anonymous referees, Professors Tsunehiro Otsuki and Fumio Ohtake, Masaru Sasaki, Yasuyuki Todo, Yasuyuki Sawada, Aya Suzuki, Petr Matous, Tomoharu Mori, Hirofumi Kurokawa, Takeshi Aida and the participants at Tokyo Workshop on International Development, Singapore Economic Review Conference 2015, Japanese Economic Association spring meeting 2015, and the seminars at Osaka University for their helpful comments and discussions. We would also like to thank Meas Pyseth and Vichet Sorn for providing us with useful information on rice farming in Cambodia. Financial support by the Grant-in-Aid for Scientific Research (No. 25101003) from the Japan Society for the Promotion of Science is gratefully acknowledged. All remaining errors are our own.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Whether such target incomes can be found in an agricultural context has not been formally examined. However, studies have provided evidence of the existence of target incomes in a variety of settings (Camerer, Babcock, Loewenstein, & Thaler, 1997; Farber, 2008; Fehr & Goette, 2007).

2. More details on key farmers in the project are provided by Yamada et al. (2015).

3. Farmers usually use a weighing scale at the time of selling their harvested rice because they know that they can measure the weight of rice more precisely by using the weighting scale provided by IRRI than that provided by the traders. At this stage, risk preferences do not play an important role in the adoption of weighting scale since famers do not encounter external risk and uncertainty. As for thermometers, farmers can use them to monitor grain temperature in drying grains. Since overheating of grains results in low quality milling, grain temperature is also an important indicator. If risk, uncertainty and loss related to the temperature of grains depend on whether farmers adopt the thermometer to measure the temperature of grains, risk preference will matter for the adoption of a thermometer. In our surveyed area, farmers usually spread grains in the sun (sun drying) for reducing the moisture content of grains, and it is difficult for farmers to control temperature. Thus, even if farmers use a thermometer, risk preferences do not also play an important role in the adoption of a thermometer.

4. There are no formal estimates on the extent of farmers’ losses due to inappropriate seed management in the study area. However, Hodges, Buzby, and Bennett (2011) suggest that in South Asia, losses resulting from sun drying and open storage systems represent 3–5 per cent and 5–10 per cent, respectively.

5. Our major concern is whether respondents understand the hypothetical game structure because if they do not understand the game, noises will be potentially influential to the estimates of their risk preferences. To avoid adding such noises to the estimates, we drop the observations of respondents who did not switch from Game A to Game B in all three series of the games.

6. We checked whether these 142 farmers are statistically similar to the other farmers in the original sample. In terms of some variables (such as the degree of risk-seeking, the probability weighting parameter, respondents’ sex, number of family members, and total income), there is a statistically significant difference between the 142 farmers and the other farmers. The reason for the difference is likely that our sample consists of decision-makers that are in charge of seed selection for the next cropping season and who are more likely to be either the head of the household or his wife.

7. As of 31 December 2012, US$1 was equivalent to 3909.4 Cambodian riel.

8. Camerer and Hogarth (1999) suggest that financial incentives improve the quality of measures of individuals’ risk preferences. In fact, Anderson and Mellor (2009) measure risk preferences by employing two types of elicitation methods, an economic experiment with financial incentives and questions on hypothetical gambles, and investigate the stability of individuals’ risk preferences. The researchers find that for most individuals, risk preferences are not stable across elicitation methods. However, the researchers show that measures of individuals’ risk preferences for some individuals are strongly correlated across elicitation methods. The researchers’ results suggest that unobserved subject characteristics such as comprehension or effort affect the quality of risk preference measures. In our questions, enumerators explained the structure of our game using picture cards until the subjects understood the structure of the game (see Appendix A). In addition, we dropped observations of respondents who did not switch from Game A to Game B in all three series of games in the analysis, because they may not understand the hypothetical game. The procedure improves the quality of our measure of risk preferences.

9. Previous studies indicate that land tenancy arrangements or land tenure status may affect agricultural investment (for example, Ali, Abdulai, & Goetz, 2012; Soule, Tegene, & Wiebe, 2000). In the context of our study, land tenure status may affect the adoption of moisture meters. However, in the area that we surveyed, nearly all of the farmers were owner-cultivators. Therefore, we do not include a variable representing land tenure status in Equation (3)(3) $Moistur{e}_{iv}={\delta }_0+{\delta }_1{\sigma }_{iv}+{\delta }_2{\lambda }_{iv}+{\delta }_3{\alpha }_{iv}+\xi {\mathbf{X}}_{\mathbf{i}\mathbf{v}}+{\mu }_p+u_{iv}$(3) .

10. In the area we surveyed, all farmers knew the key farmer, and the share of farmers who had only observed the key farmer (without talking to him) was only 2.7 per cent. We therefore combined the category of farmers who had only observed the key farmer with the category of farmers who had talked to the key farmer and used the combined categories as the reference category in the analysis.

Additional information

Funding

This work was supported by the Japan Society for the Promotion of Science [Grant number 25101003].