Risk, time and adoption of perennial energy crops: the French case.

Summary The objective of this thesis is to identifier the determinants of farmers' adoption of perennial energy crops by focusing on issues of risk and time. The analysis is based on the case of miscanthus (Miscanthus giganteus) and switchgrass (Panicum virgatum) in France, but is potentially generalizable to other perennial crops. The relevant public and private decisions to encourage the development of these crops are debated. In the first paper, we evaluate the impact of risk and time on the optimal area of miscanthus and switchgrass in a field crop farm in the Centre region. We go beyond the usual net present value calculation by explicitly accounting for risk aversion and intertemporal fluctuations through the expected utility and discounted utility frameworks. Our results show that the two perennials are on average less profitable than the traditional rapeseed/wheat/barley rotation. However, they can be very competitive as diversification crops when suitable production contracts are offered to farmers. The second, third, and fourth articles exploit data from a survey and experiment conducted with 111 farmers in Burgundy who were faced with the choice of whether or not to grow miscanthus. The second article describes miscanthus production in Burgundy. We show on the one hand that miscanthus is mostly planted on marginal plots that are not very profitable for traditional uses. We also show that, even in the presence of long-term contracts, farmers perceive miscanthus as less risky overall than wheat, but remain concerned about specific risks with unlikely but extremely unfavorable outcomes. In the third paper, we estimate farmers' risk preferences from the experimental data. We apply a structural estimation method to a decision model consistent with prospect theory. We then review a number of implications of this theoretical framework for agricultural economists. Our estimates indicate that prospect theory explains our data better than standard expected utility theory. This is because farmers are loss averse and distort probabilities to give significant weight to extreme events. In the fourth paper, we examine the relationship between miscanthus adoption and farmer and farm characteristics, in particular individual risk and time preferences. The latter are represented by experimental measures obtained under prospect theory and hyperbolic discounting respectively. Our results suggest that the probability of miscanthus adoption depends on the degree of loss aversion of farmers and the extent to which they distort probabilities. However, the impact of these two factors varies with the type of plot considered and the farmers' reference point. In addition, the probability of adoption is higher the higher the proportion of unprofitable land on the farm, and the higher the proportion of land that is not already being used for livestock production.