The Effect of Insurance on Farmers' Production, Technical Efficiency and Input Use: An Endogenous Stochastic Frontier Model to Analyse the Italian Case

The presence of risk in the agricultural sector has important implications for production decision-making and, therefore, the economic and environmental performance of farms. Given the risky environment, the EU policymakers introduced several risk management tools aimed at the reduction of the income variability of farms. The adoption of risk management tools impacts production decision-making in several ways. Therefore, analysing the interrelation between farm performances and risk management tools is crucial for policymakers to foster both the economic and environmental sustainability of farming. Despite the significant role of risk and risk management tools in agricultural production decisions, most studies on farm-level performance analysis do not account for them. Moreover, when included, it is usually assumed that risk and risk management tools are independent of the error terms in the model. Therefore, endogeneity issues are often not considered, which could overstate or understate the effect of risk and risk management tools on farm performance. Indeed, producers may modify input use in response to observed adverse events resulting in a correlation between inputs and statistical error. In addition, endogeneity may arise when including the risk management tools. The model misspecifications due to the absence of endogeneity treatment may lead to erroneous inferences about the estimates of input elasticities, economies of scale, and inaccurate estimates of technical efficiency. Consequently, the analysis may lead to incorrect interpretation and, ultimately, to wrong policy advice. Among the risk management tools, crop insurance represented the most funded instrument in the EU. In spite of recent growth in the scientific literature on crop insurance in agricultural economics, only a few studies have concentrated on the effects of crop insurance expenditure on farm outcomes. Specifically, the impact of insurance on productivity and technical efficiency has received scant consideration. Therefore, the general objective of this thesis is to quantitatively assess the effect of crop insurance on input use, productivity, and technical efficiency of Italian farming producers. In particular, this thesis examines whether insurance adoption could reduce farmers' inefficient input use that results from the uncertainty of the results. Grape farming has been selected as a case study since it is the sector where crop insurance has been most adopted in Italy. The novelty of this research relates to the inclusion of crop insurance adoption endogeneity into the stochastic frontier approach, allowing for the estimation of parameters with a higher degree of accuracy. The thesis is divided into seven chapters. Each chapter attempts to lay the groundwork for answering the research question. In particular, Chapter 1 introduces the background, the problem, and the research question, while Chapter 2 introduces the Italian crop insurance regulations and the spread of subsidized crop insurance in the grape-producing sector. Chapter 3 describes the theoretical background. Specifically, the production function is introduced, which is the first step to comparing the performance of producers. Subsequently, productivity, technical efficiency, and their difference are discussed. Following that, the optimal input use and the effects of risk-aversion on production choice have been discussed. Finally, the impact of insurance on farm performance has been reviewed. In particular, the dilemma which regards the re-optimization or the moral hazard effects arising from the insurance adoption has been investigated. Then, Chapter 4 is based on an in-depth literature review on farm productivity and efficiency, which has been conducted using a scoping review methodology, focusing on studies that have included risk and risk management tools within the stochastic frontier analysis. The main contribution of the review relates to the indication of a literature gap concerning studies accounting for endogeneity and the clarification of methods used to account for it by using a risk-accommodating stochastic frontier approach. Despite the increasing methodologies proposed in the literature to deal with endogeneity, only a few studies have treated it in farm risk-performance evaluations when using the stochastic frontier analysis. According to the findings of the review presented in Chapter 4, it can be concluded that there is a literature gap regarding the adoption of a comprehensive approach capable of dealing with endogeneity when assessing farm productivity/technical efficiency and risk. Neglecting endogeneity in these analyses may lead to biased estimates and, thus, distorted policy recommendations. Endogeneity and risk issues need to be concurrently addressed to make strides in achieving economic and environmental sustainability. The comprehensive approach could help to achieve more accurate estimates that could yield recommendations that ensure improved productivity and technical efficiency of farmers. Bearing in mind that, a case study has been implemented to assess how insurance affects Italian specialized-quality grape growers' production, technical efficiency, and input use while accounting for the endogeneity of the crop insurance adoption. Therefore, a panel instrumental variable stochastic frontier approach is applied over the years from 2008 to 2017 using data from the Farm Accountancy Data Network. The methodology, the econometric strategy to deal with endogeneity, the dataset, and the model specification are presented in Chapter 5, while the case study is documented in Chapter 6. The findings highlight the need to account for endogeneity brought on by the adoption of insurance. Moreover, it was found that crop insurance increases output and efficiency while decreasing the need for intermediate inputs in Italian grape farming. It implies that insurance assists in reducing the suboptimal input use caused by risk-aversion and the uncertainty of farming outcomes. Finally, Chapter 7 discuss the results, summarizes the main conclusion of the thesis, and highlights the limitations and future research directions.