120 likes | 293 Views
Value of Information and Value of Control in fisheries management: North Sea herring as an example. Samu Mäntyniemi, Sakari Kuikka, Laurence Kell, Mika Rahikainen and Veijo Kaitala. ICES Annual Science Conference, Halifax, Canada, Sep 26 th , 2008. Outline. Loose definition Conclusions
E N D
Value of Information and Value of Controlin fisheries management: North Sea herring as an example Samu Mäntyniemi, Sakari Kuikka, Laurence Kell, Mika Rahikainen and Veijo Kaitala ICES Annual Science Conference, Halifax, Canada, Sep 26th, 2008
Outline • Loose definition • Conclusions • Application to North Sea herring population
Value of information (VoI) is the amount a decision maker would be willing to pay for information prior to making a decision.
Value of information • Can be calculated for information sources • e.g. gathering of monitoring data • Expected increase in expected profit when using the information source • Or maximum price to pay for the information source • Or expected loss if you ignore the information source • Leave existing knowledge out from stock assessment -> less expected profit from fishery • Emphasizes the importance of prior information • Used widely in other areas outside fisheries
Properties of Value of Information (VoI) • VoI depends on the initial amount of information about the state of nature • VoI = 0, when the state of nature is already known exactly • VoI is highest when the existing knowledge is poor • If new information changes the optimal decision, then VoI > 0
Conclusions I • Value of information can be used to prioritize research activities for management needs • For each information source, calculate: Value of Information – Cost of information • Then rank the sources • If the cost is higher than the value, do not buy the information
Conclusions II • Could be utilised at different levels • Fishing companies: plan investments to technology • Managers: plan investments on new research • EU commission: planning of research agenda • Prerequisites • Probabilistic formulation of information • Bayesian inference in system assessment • Numerically stated management objectives • Mix of economic and social objectives = difficult • Clearly defined set of alternative management actions • Not possible to define VoI for research that may provide completely new hypotheses!
Example: North Sea herring • Objective: maximize expected profits over next 20 years • Herring price and fishing cost assumptions from Rahikainen et al. 2008 (Presentation nr:o 7, 30 minutes ago) • Alternative decisions • Increase or decrease the current fishing mortality and keep it constant over the 20 year period. • Uncertainty about • Natural mortality, fishing mortality, selection curve, type of stock-recruitment (SR) relationship, SR-parameters, true catches at age and about the true status of the stock • How much to pay for perfect information about the type of SR relationship?
Bayesian stock assessment model • Age-structured, models full life cycle • Includes both Beverton-Holt and Ricker stock recruitment relationships as hypotheses • Fitted to the North Sea data set used by ICES Herring Assessment Working group • Output: posterior probabilities for Beverton-Holt and Ricker models • Ricker : 0.57 • Beverton-Holt : 0.43
Value of information B-H=235M NOK IF B-H Ricker=243M NOK Current knowledge Bayesian stock assessment: P(B-H | data)=0.43 P(Ricker | data)=0.57 Currently optimal IF Ricker Value of Information = 0.43 x 235 + 0.57 x 243 = 240M NOK Optimal if SRR = Ricker Optimal if SRR = B-H
Decide about investing • If perfect knowledge about the type of SRR costs less than 240M NOK, go and buy it! • Remarks • Perfect knowledge impossible • Calculate value of imperfect knowledge instead • Need knowledge about the precision of the research • More complex, but possible to do • Value of overconfidence? • Act as if B-H or Ricker was true, when actually uncertain • Reverse the concept of value of information -> loss because of overconfidence -> if forced to, choose to use the one with less loss