Adverse shocks might have a direct impact on the production of rural households by destroying output and physical assets.They might also have an indirect effect by altering farmers’ behavior towards risks.Under dysfunctional and flawed insurance markets, rural households in developing countries have become more risk-averse after experiencing co-variate and idiosyncratic shocks.However, just a few studies take shock experience and farmers’ risk attitude in examining their impacts on crop production.While these previous studies provide important insight, there are a number of research gaps that need further investigation.First, the endogeneity of risk aversion has not been addressed.Second, while rural households in developing countries have to cope with a wide range of shocks and production risks, previous studies mainly considered droughts and crop pests in the analysis disregarding other shocks such as floods, storms, and diseases.Third, previous studies did not examine how changes in farmers’ risk attitude impact farming efficiency to validate whether farmers’ application of pesticides and fertilizers is efficient, especially for risk-averse farmers.Against this background, we use a panel dataset collected in Thailand to examine the impacts of risk attitudes on fertilizer and pesticide use, and investigate the effect of adverse shocks and risk attitudes on technical efficiency in rice production.Thailand is relevant because agricultural production plays an important role in its rural economy.Addressing these research questions is necessary for policy responses to the harmful impacts of the inefficient application of synthetic fertilizers and agrochemicals on rural households’ production and the environment.The rest of the paper is as follows.Section 2 reviews the literature.Section 3 introduces the study sites and data.Section 4 describes the methods for data analysis.Section 5 discusses the findings.Section 6 concludes with policy recommendations.Although the relationship between risk attitude and input application has been examined in a few studies,dutch bucket hydroponic the findings on the roles of pesticides and fertilizers show mixed directions.
However, they could also play a risk decreasing role.For instance, Rajsic et al.found that nitrogen was a risk‐increasing input, implying that risk‐averse farmers tend to apply less nitrogen.This finding is supported by Möhring et al..On the contrary, Khor et al.stated that less wealthy farmers had a lower level of fertilizer use when their risk aversion increased.This finding aligns with Salazar and Rand that fertilizers are risk-decreasing inputs.Farmers who are more unwilling to take risks might overuse fertilizers because they think the crops need an additional amount of fertilizers.With regard to pesticides, a key motivation behind the application of pesticides is to provide a means of insurance against yield losses/damages caused by pests and diseases.These studies revealed that the higher the degree of uncertainty regarding pests’ damages, the higher the volume of pesticide application, despite any given levels of pest infestation and pesticide costs.Liu and Huang confirmed the risk-reducing role of pesticides.Nevertheless, pesticides could also play a risk-increasing role.Möhring et al.pointed out that risk attitudes affect differently on pesticide use depending on the types of pesticides.Recently, Salazar and Rand examined the impacts of production risks on pesticide use and concluded that pesticides are risk increasing inputs when more risk-averse rice producers apply fewer pesticides.Although these previous studies provide important insight on the association between risk attitude and input application, there are a number of research gaps that need further investigation.First, farmers in developing countries live in a highly vulnerable environment with a wide range of adverse shocks.However, only a few studies simultaneously take these aspects into account when estimating the impact of risk attitude on crop production.Rural households’ behavior under risks might explain low agricultural productivity, vicious cycles of poverty, and determination of risk-aversion in the loss domain to maximize investment decisions.Uncertainties caused by adverse shocks affect rural households’ risk attitudes that might lead to improper applications of inputs and, therefore, reduce technical efficiency.In this case, their fear of uncertainties may encourage them to apply more inputs than efficient levels, and this overuse is wasteful and harmful for the environment and their health.As a result, farmers with high levels of risk aversion could culminate in economic decisions that lead to relatively less income.Thus, accounting for diverse shock types in estimating input application still deserves further attention.Second, farmer’s risk attitude is endogenous.There is a significant and robust linkage between risk aversion and wealth levels in the form of income or assets of the households.
Farmers’ risk attitude can also be affected by household characteristics such as age, education, and gender.Externalities can further influence the risk aversion of rural households in the form of adverse shocks.Therefore, estimations of input use and risk preferences ignoring these aspects might produce biased results due to the problem of endogeneity.Third, farmers’ risk aversion might change overtime; however, most previous studies on risk attitude and input application in developing countries relied on cross-sectional data because long-term panel data with information on risk aversion might not be available.Thus, using panel data for this type of study is relevant to produce more reliable evidence since it allows to control for unobserved sources of heterogeneity.Hence, our study contributes to filling these research gaps.We simultaneously examine the impact of risk attitudes and shocks on input application and technical efficiency in rice production.By employing a balanced panel dataset of rice producers in Thailand, we first investigate the association between risk attitude and input use in the context of shocks.We control for the potential endogeneity of risk attitude by employing an instrumental variable regression.Then, we estimate the technical efficiency in rice production through a stochastic frontier model for panel data proposed by Greene to justify the effects of improper input application caused by farmers’ risk attitudes and shocks.One of the advantages of this model is that it allows us to estimate time-variant efficiency and can distinguish the unobserved heterogeneity from the inefficiency component.The findings are expected to enrich the literature on risk attitude and chemical input application and provide useful insight for formulating public policies to mitigate the negative impacts of shocks, improve production efficiency, and reduce the harmful effects of chemical overuse on the environment.Data for this research are from the “Poverty dynamics and sustainable development: A long-term panel project in Thailand and Vietnam ”, funded by the German Research Foundation.This project aims to generate a better and in-depth understanding of income and vulnerability to poverty dynamics in rural regions of the emerging economies of Thailand and Vietnam.Following the guidelines of the Department of Economic and Social Affairs of the United Nations , the sampling process included a three-stage stratified random sampling procedure based on the administrative system of each country.In Thailand, the survey was conducted in three provinces, namely Buriram, Nakhon Phanom, and Ubon Ratchathani , where majority of the households live in rural area and are dependant on agriculture for their livelihood.In the first stage, sub-districts were selected in each province.Then, two villages were chosen with a probability proportional to the size of the population.At the third stage, a random selection of ten households was made based on the list of all households in the sampled villages with equal probability,Klasen and Waibel for detailed information of the survey’s designation and implementation.
For this research, we use a balanced panel of 1220 rice farmers collected in 2013 and 2017.In this survey, the information of risk attitude is a self-assessment scale similar to the one in the German Socioeconomic Panel conducted by the German Institute for Economic Research.In this self-assessment, the respondents were asked to self-evaluate their risk attitude on a shown scale ranging from zero to ten.Although this kind of self-assessment might not perfectly reflects risk attitude, it has been validated as an appropriate indicator for respondents’ risk preferences and has been widely applied in studies on risk preferences.With regard to shock experience, the respondents were asked to report shock events that they experienced in the reference period “Was your household affected by any of the following [events] between 1st May 20XX to 30th April 20XX”.The length of the reference period was defined by the gap between the current and previous waves.In this research, we focus on weather shocks , crop pests and diseases.We take the respondents’ exposure to shocks in the last 12 months into account as indicators of shock impacts such as production costs, yield, and efficiency are based on a 12-month recall period.We prevent misreported shocks of respondents by cross-checking between reported shocks and their losses due to the events.Then, we generate a dummy variable of households who are exposed to weather shocks,dutch buckets system crop pests and diseases.These reported shocks are strongly relevant to agricultural production in rural areas in developing countries.In the TVSEP data, input costs are recorded with a wide range of cost categories such as land preparation, seedling, weeding, fertilizers, pesticides, irrigation, harvest costs, and other costs.The other costs include additional costs that do not fit any in the listed cost categories, for example, of pre-processing before selling.This study uses fertilizer volume, fertilizer expenditure, and pesticide expenditure as key variables to analyse the impacts of farmers’ risk attitudes on input applications.We use the expenditure on pesticides instead of quantity use because the data do not record the amount of pesticides.We control for price differences by using constant monetary values adjusted to 2005 prices.Besides key variables, namely farmers’ risk attitudes, rice production, and shocks, we control for other characteristics of rice farm households such as household’s demographic characteristics, farming characteristics, physical capital, and village characteristics.Table 1 provides a descriptive summary of the data.The descriptive statistics show significant differences in rice output, expenditures on fertilizers, pesticides, seedling, weeding, irrigation, and other costs, but not the fertilizer quantity, land preparation costs, and harvest costs between 2013 and 2017.While the use of inputs is higher, the rice productivity was lower in 2013 than in 2017.
The average farming area of rice farmers in Thailand is about 3.24 hectares , and approximately two household labourers engage in farming activities.The experience of shocks appears to be different over time.Particularly, farmers reported more weather shocks in 2013 but almost the same level of crop pests in 2013 and 2017.Overall, farmers who experience shocks appear to significantly have lower rice yield, lower expenditure on land preparation, higher expenditure on fertilizers, pesticides, seedling, and other costs, while fertilizer use and expenditures on weeding, irrigation, and harvest are not significantly different.Households experiencing shocks have larger farming areas and more household members engaging in agriculture than non-shock households.Households with shock experience also tend to have a lower level of willingness-to-take risks than the households without shock experience.Table 2 shows the demographic characteristics, farming characteristics, physical capital, and village characteristics of rice farmers in Thailand.The average age of the households’ head is about 60 years old with around five years of schooling.The household size and dependency ratio are significantly different both between 2013 and 2017 and between shock and non-shock groups.On average, rice farm households in Thailand have about five members.The average distance from farmers’ house to all land plots is 2.23 km.The village characteristics show that the vast majority of households in rural Thailand have access to electricity , but only a small percentage of them have cable internet at home.The instrumented risk attitude variable shows a negative impact on input applications with a significance at less than 10% level.This implies that both fertilizers and pesticides can be considered risk-reducing inputs in rice production in Thailand.The estimations of fertilizer use in both quantity and monetary values show almost the same effect of farmers’ risk attitudes on the application of fertilizers.In other words, the more the farmers avoid risks, the more they apply fertilizers and pesticides.This also points out that becoming more risk-averse influences them to apply more inputs, even though these applications are improper.Our results remain consistent with lagged values of risk attitudes from the previous waves.Compared with a similar rice exporting country, our results of the correlations between risk attitude and input use support the findings from Salazar and Rand that fertilizers are risk-decreasing inputs in Vietnam, but pesticides have an opposite role.This difference can be because of the intensive level in rice production between the two countries or the biased results from the endogeneity problem unaddressed in their estimation.In short, uncertainties motivate rice farmers to use more fertilizers to enhance crops production because of their aversion behavior to losses.Besides, Salazar and Rand found that droughts negatively affect pesticides use.This is contrary to our findings.