Browsing by Author "Jaana, Haavisto"

Policy makers must make decisions regarding budget allocation between policies and research. Only actions improve the state of the system, but knowledge increases the probability of effective action. The outcome of environmental policies is usually uncertain, and the question remains: should we invest more in research or use resources for additional policies? Under uncertain decision making, it is clear, that doing and knowing go hand in hand. Still, there is a lack of scientific analyses about the relationship between these features. This paper analyses how uncertainty in policy implementation outcome (i.e., Value- of- control, VoC) effects the need for additional knowledge, which is measured using value- of- information (VoI) analysis. Additionally, the paper analyses how the obtained results can help in the allocation of resources between policy and research. To answer these questions, I use a published Bayesian decision model by Helle et al. (2015) as a source for further analysis. It is an influence diagram model, consisting of two decision variables, 40 random variables and 13 utility variables which are described in euros, allowing the monetary summarisation of utility. I introduce levels of implementation uncertainty to the other decision by placing an additional random variable to influence the successor variables of the decision. I define the levels of uncertainty using two co-variation methods, first one being the proportional co-variation method, and the second one the order- preserving uniform method. By this way, I analyse and compare the effect of distinct levels of decreased controllability (VoC) to VoI analysis results. These two methods describe two alternative ways of modelling the uncertainty in implementation, which is always uncertain when we consider future actions that have yet to been implemented. I conduct the analysis separately for both co-variation approaches, and 10 alternative levels of implementation uncertainty, to enable systematic comparability between the chosen methods, and to learn alternative ways to consider the relationship of controllability and knowledge. First, I preform the VoI analysis only for the policy that is subjected to implementation uncertainty. Secondly, the analysis is done for both decision variables and the Single Policy Updating (SPU) algorithm of Hugin software is used for detecting optimal policies for different implementation uncertainty levels. In other words, I show how the various levels of controllability of the system impact the needs to carry out research. I argue, that this is a fundamental question for many environmental policy questions, such as climate change, eutrophication, loss of biodiversity and in my example, risks of oil spills. The results show a consistent, but interesting effect of decreased controllability to the VoI analysis results. Increase in implementation uncertainty raises the overall VoI and increases the number of variables presented with VoI, i.e., once the estimated uncertainty of controllability increases, the chance to achieve desired results increases only by knowing more. When only one decision variable is included, VoI increases to the point of no control, indicating that VoI is zero when controllability of the system is zero, indicating that there is no point of carrying out research, if the knowledge cannot be used to improve the effects of actions. When both decision variables are included in this case, VoI increases to a certain point and decreases after that. This study highlights the need for such analyses in decision problems, where uncertainty in policy implementation is often overlooked. This is the case with most deterministic, point estimate models. I argue, that this type of analysis would lead to more effective solving of environmental problems.