Risk aversion hinders forestry professionals to adapt to climate change

Abstract

Forestry professionals’ attitudes towards risk and uncertainty under climate change, together with their perception about suitability of adaptation strategies, were investigated in Central Europe. We applied an original methodology based on lottery choices to quantify their risk and uncertainty attitude, combined with a questionnaire study about the adaptation to climate change. We tested the hypothesis that the higher the risk and uncertainty aversion of respondents, the higher the trend towards changing business-as-usual to adaptive decisions. The result falsifies the research hypothesis since uncertainty aversion has no effect, whereas risk aversion has a negative impact on the decision to adapt. We argue that risk of any change in the business-as-usual is higher than the expected risk of climate change impact in Central European forestry. We conclude that having access to novel information about expected impacts of climate change, taking into account institutional challenges and supportive experiences about useful adaptation strategies, may convince forestry professionals to adapt in the future.

Appendices

Appendix 1. The lottery choices

Table 4 presents the ten lottery choices proposed to the forestry professionals in a risk context. We used exactly the same tables (payoffs, probabilities, bounds for risk aversion) as Reynaud and Couture (2012). For example, decision 1 corresponds to a choice between Option 1: a probability of 90% to have a payoff of €16 and 10% chance to have €20, and Option 2: 90% chance to have €1 and 10% chance to have €38.5.

Table 4 The ten lottery choices of the MPL procedure under risk

Table 5 The ten lottery choices of the MPL procedure under uncertainty

Table 6 Characterization of the preferences

Table 5 presents the ten lottery choices proposed to the forestry professionals in an uncertain context. For example, decision 1 corresponds to a choice between Option A: 50% chance to have a payoff of €35 and 50% chance to have €0, and Option B: having €25 or €0 but the associated probabilities are unknown.

The measure of the preferences corresponds to the number of safe choices to characterize risk aversion and to the number of risky choices to elicit uncertainty aversion. Table 6 presents the classification in terms of attitude towards risk and uncertainty function of the number of safe and risky choices realized, respectively.

Looking at the distribution of the numbers of safe and risky choices reveals that the higher numbers of safe and risky choices are for the neutrality threshold, respectively 4 and 5. Another observation is that 0 and 1 safe or risky choices are very rarely selected by forestry professionals so that extreme loving is not really represented in the sample. Conversely, 9 and 10 safe choices are chosen each by approximately 5% of the sample indicating extreme risk aversion. In the same vein, 9 and 10 risky choices are selected each by approximately 10% of the sample indicating extreme uncertainty aversion.

Appendix 2. The cross strategies analysis

Table 7 proposes to cross the already implemented strategies (in rows) with those planned in the near future (in columns). The letters “a” to “m” correspond to the 13 strategies proposed to the forestry professionals with: a = Plant genetically modified species, b = Assisted in tree regeneration, c = More species mixture, d = Less species mixture, e = More forest fertilization, f = Less forest fertilization, g = Longer rotation, h = Shorter rotation, i = Heavy thinning, j = Light thinning, k = More forest insurance, l = Less forest insurance, m = Other measures.

Table 7 Crossed analysis of strategies already implemented and those planned (in number)

We observe that when a strategy has already been implemented in the past (already adapted), most of time the foresters plan to continue it in the future (diagonal of italicized values). The first italicized value means then that 4 forestry professionals implemented adaptation strategy a (Plant genetically modified species) and plan to continue it in the future. In this sense, we observe that the choice of a strategy in the past is positively and significantly correlated with the choice of the same strategy in the future. For example, the choice of strategy b in the past and the choice of strategy b in the future are positively and significantly correlated at 1% (Pearson correlation coefficient: 0.578; p = 0.000). The same result applies for each strategy, except strategy c (Pearson correlation coefficient: 0.185; p = 0.126). However, sometimes, the forestry professional complements this first phase of adaptation with a new one, composed with different strategies. As a consequence, we observe that combinations of strategies emerge. For example, professionals having adopted strategy c often planned to adopt strategy b. Strategies b and c are those that have been selected the most by respondents and, they are also the strategies that have been most of time combined between them, but also with others, in particular with strategies g, h and i.

Appendix 3. Full detailed results of the two logit regressions

Table 8 presents the full results of the Model 1: logit model with a fixed effect specific to adaptation strategies and a random effect per individual, and of Model 2: logit model with only a fixed effect specific to adaptation strategies. As the random effect is not significant in Model 1 (LR test on absence of random individual effect: chibar2(01) = 1.02; Prob ≥ chibar2 = 0.1561) and as the results are similar to Model 2, we decided to present the results of Model 2 in the text. In addition, the significance of the variable in the model without individual effects increases. We found the significant variables appearing in Table 2 but also other ones. The variables called « Strat » with an associated letter correspond to the 11 adaptation strategies (13 proposed minus 2 never selected) minus the 11th (Strat_k is the less selected strategy, i.e. more forest insurance). The impact of these variables is always positive, meaning that the probability of the chosen strategy is always higher than choosing Strat_k (i.e., strategy to which the others are compared). The significant positive coefficient on Strat_a, Strat_b, Strat_c, Strat_g, Strat_h, Strat_i, and Strat_m means that these strategies are selected significantly more frequently than the Strat_k. They correspond to strategies that are more frequently selected by forestry professionals, and then, they significantly contribute to the probability of adaptation.

Table 8 Estimations of the parameters for the two logit models

We also ran a simple logit regression by adaptation strategy.

Such a logit may be performed for the strategy « Assist in tree regeneration » because this strategy is well balanced, i.e. 45 forestry professionals had already implemented it and 43 had not. The results reveal that some significant variables are the same, such as Income (parameter value = 0.724; p = 0.077) and French (parameter value = − 2.269; p = 0.006), while other appeared such as Dominant_fir (parameter value = − 2.179; p = 0.019), Private (parameter value = 2.631; p = 0.023), More_coniferous (parameter value = 2.431; p = 0.015). This means that having fir as the dominant tree species reduces the probability of assistance in tree regeneration while managing private forest and having more coniferous than broadleaved increases the propension to assist in tree regeneration.

In the same vein, a logit regression may be performed for the strategy « Heavy thinning » since it had already been implemented by 28 forestry professionals (and 60 had not implemented it). We show that the variable Income (parameter value = 0.396; p = 0.082) has a significant and positive effect on the probability to implement heavy thinning to face climate change. The variables Dominant_fir (parameter value = 0.908; p = 0.097) and Hunt_objective (parameter value = 1.645; p = 0.010) indicate that the probability to adapt is higher if hunting is one of the professional’s main objectives and if fir is the dominant tree species.

Appendix 4. The questionnaire