FORESTry policies design for under climate change: improving Time and RIsk Preferences measures
PI : Antoine Leblois (UMR 356 Laboratoire d’Economie Forestière – LEF)
________________________________________________
Context — Recurrent forest fires, heatwaves (2003) and windstorms (1999, 2003) in France are striking examples showing that forest resource is subject to major climate hasards and especially extreme climate events (low probability / high damage). Those natural hasards lead to large scale damage and are highly unpredictable, creating a need for new and well designed public policy. Besides, if adaptation of agriculture to climate change has been extensively studied by economists, in particular with a behavioural approach, forest management has been less focused on. The literature on agriculture has already found evidence of an impact of individual risk preferences on technology adoption (Liu, 2013) and input use (Liu and Huang, 2013). It is also the case for insurance demand (Petraud et al., 2014; Babcock, 2015; Piet et al., 2016). Better knowledge of individuals preferences using experimental economics thus may help designing public policies and foster insurance demand (cf. Jaspersen (2015) for a review on hypothetical insurance adoption). Forest management modelers have underlined the importance of eliciting risk and time preferences parameters of forest managers (Knapp and Olson, 1996; Kangas, 1994; Dragicevic et al., 2016), but literature about such parameters is limited (Lönnstedt and Svensson, 2000; Stenger, 2009; Andersson and Gong, 2010; Brunette et al., 2014b). Thus far, models of ecosystem dynamics for decision analysis have not been able to predict the effect of different management prescriptions as required for adaptive forest management (Yousefpour et al., 2012).
Objectives — The high diversity of observed decisions about forest estate management when facing natural hasard is conditioned by forest intrinsic characteristics: physical growth, environment, operating and management costs, high potential diversification within forest assets (Dragicevic et al., 2016) or among other capitalized assets and long time horizon. Growth of trees can indeed exceed 50 or 100 years, depending on species, obliging managers to foresee long run evolutions of their environment (of ecosystems, institutional constraints etc.) and apprehend extreme events. It thus motivates specific analysis of time perception since it mobilizes very long run anticipations. Forest management related decisions are indeed largely conditioned by attitudes towards risk and time (Koskela and Ollikainen, 1999; Peltola and Knapp, 2001; Couture and Reynaud, 2008; Sullivan et al., 2010). Those attitudes are synthesized into risk aversion (and potentially loss aversion specific parameter and probability weighting) and time discounting parameters of the utility function. Those parameters are generally calibrated using values found in the general economic literature, 2arbitrarily fixed or results robustness tested using sensitivity analysis. However, it is possible that forest owners and managers have heterogeneous time and risk preferences [1], inducing a bias in the analysis. A gap our project will try to fill, by eliciting individual preferences through an experiment. There is particularly limited empirical evidence about forest managers’ behaviours towards natural hasards in France. A first study is a survey about private forest owning (“Structure de la propriété forestière privée en 1999”. Agreste no 144. Novembre 2002), with too limited scope for allowing econometric analysis. A second one shows that forest owners are averse to natural hasards but does not consider heterogeneity of such preferences in the population Stenger (2009). Brunette et al. (2014b) shows that risk aversion of forest owners is high and that it is linked with the gender and age, as well as the share of forest estate in their patrimony. Moreover, the authors show that the chosen rotation periods are influenced by behaviours towards risk. The above-mentioned studies are limited to expected utility theory (EUT), that has been challenged by different paradoxes [2]. The EUT (Neumann et al., 1944), with a single parameter of risk aversion, has been improved with loss specific aversion (PT, (Kahneman and Tversky, 1979)) and probability weighting (including rank dependent utility, Quiggin (1982)). The cumulative prospect theory (CPT, Tversky and Kahneman (1992)) including both characteristics. CPT, including probability weighting, one of the most robust finding of the last 30 years (Barberis, 2013b), in a unified framework, is however a rather untractable empirically. It has been tested on French farmers (Bocquého et al., 2014), but not with foresters, although it allows a more precise assessment of perceptions of rare events, and thus natural disasters (Barberis, 2013a).
Expected results and impacts — The lack of data on forest managers’ preferences also limits the development of risk management (ex ante, e.g. diversification) and risk coping (ex post, e.g. insurance) tools. Public policies thus may be improved, especially in France where catastrophic events are largely managed by the state. Our work will try to go further that road, to complement previous research and notably for understanding perception of risk and time and possible interactions, considering above-mentioned limitations of the EUT framework. This project thus may help understanding the perception of rare events by incorporating probability weighting and loss specific aversion, keeping in mind the existence of a trade-off between completeness and empirical tractability of utility theory.
——————
1. In France, the large majority (about two third) of forest owners are private individuals endowed with small units of forest land (average of 2.9 hectares). They are managing those forest estates, often inherited, as a risky component of their relatively diversified patrimonial capital.
2. Allais (Allais, 1953) and Ellsberg (Ellsberg, 1961) are the most famous ones.
