sakari kuikka university of helsinki maretarium, kotka content: 1)decision making in general and in...
TRANSCRIPT
Sakari KuikkaUniversity of HelsinkiMaretarium, Kotka
Content:
1) Decision making in general and in fisheries2) Value-of-information3) Value-of-control4) Commitment: role of understandability
Use of decision analysis in the evaluation of scientific information
Main results of the talkWorld Cup Icehockey, last night
Canada – Finland 3-2 (1-1,1-1,1-0)
00.52 Joe Sakic (Mario Lemieux, Eric Brewer) 1-0 06.34 Riku Hahl (Toni Lydman, Aki Berg) 1-1
23.15 Scott Niedermayer (Kris Draper, Joe Thornton) 2-1 39.00 Tuomo Ruutu (Toni Lydman) 2-2
40.34 Shane Doan (Joe Thornton, Adam Foote) 3-2
Uncertainty Rowe (1994):
• Temporal uncertainty: future and past states• Structural uncertainty (uncertainty due to
complexity, related to control)• Metrical uncertainty (uncertainty in
measurements)• Translational uncertainty (uncertainty in
explaining uncertain results)
Bias of ICES stock assessments
Errors in assessments (1988-1999)
0
5
10
15
20
25
30
35
0 1 2 3 4 5 6 7 8 9
Ratio SSB-predicted/SSB-truth
Fre
qu
en
cy
Sparholt & Bertelsen, 2002
Part I : Decision making and decision analysis
”Predicting the outcome is far more difficult than the ranking of decision options”
Actions and Decisions
Fisheries management:
”Economically effective control of an uncertain biological system by the politically possible juridical control tools”
Only actions will increase utilities (getting closer to objectives), not predictions or scientific estimates as such
Management of environment and fisheries
1) What are your aims?2) What are your management tools3) What do you have to know to use those tools4) How do you know whether your management is worthwhile
Types of decision Analysis 1) Analysis of objectives: Analytic Hierarchy Process: AHP
= systematic weighting of objectives and their linking
to decision alternatives
2) Analysis of knowledge and actions: Decision trees and influence diagrams.
= analysis of probabilistic information in a decision framework
State of nature Knowledge
Action New state of nature
Production potential of the stock (real state of nature)
How well we can measure/assess ?= quality of the science
Availableknowledge
How strong will be the impactof decision on nature (e.g. implementation uncertainty)
= aim
Utility: dependent on action and on the real state of nature
Chain of knowledge and actions
Step 1: Decision to implement new economic subsidies to decrease the effort
” Decision to act”
Step 2: Change in fishermens behaviour
”How humans act?” Uncert: which vessels?
Step 3: Impact on nature
” How the SSB or recruitment will change”
2
3
1
4Step 4: Degree of success
”How do we valuate changes?”
Fisheries management:Chain of humans and nature
Evaluation of decision optionsUncertainties in:
a) Implementation (juridical and socio-economic part)
b) Biological impact (biological part): the gain of saving a fish
c) Current and future objectives (political/sociological part)
Lack of objectives?
Decision analysis can also show, what must the objectives be, if the available information and decisions are known: transparency
You may be able to show, that even though there are different objectives, they all favor the same decisions
=> stakeholders do not necessarily need to agree on objectives
Part II: value of knowing and value of doing: Basic elements of decision
analysis
Value of information and value of control
1) How much I should pay for the better information? = value-of-information - dependent on e.g. how much decision could change, if new
information is obtained, and how well the new decision can be implemented?
2) How much I should pay for the better control (management) of the system?
= value of control
- how much the expected state of the system could be improved, if the precision of the control would be improved
Value of Information and Control
• Expected Value of Perfect Information (EVPI): new information => choosing a different action with better outcome => information had some value
(dependent on the controllability)
• Value of Control: ability to change the value of a previously uncontrollable variable or improving of controllability (better adjustment of the system)
= Numerical estimates of key elements in the planning of control and information system (monitoring + studies)
Simplified example
Value of information: better estimate for M + decreased F => higher yield per recruit
Value of control: adjustment of M through multispecies context => higher yield per recruit
VOI and VOC
M = .2
M = .4
Example:Value-of-information
If fishing mortality of 0.5 produces catch of 2 million during the Next 20 years, and mortality of 0.7 produces 1.5 million, the information that switched the decision to 0.5 had a value of 0.5 million fish
However, expected value of perfect information EVPI (e.g. Clemen, 1996) is often estimated in advance: the likelihood of future information (study results) under various scenarios must be evaluated
The most useful studies have a high value-of-information.
The best management schemes have low estimates for the value-of-information = information robustness
0
0.005
0.01
0.015
0.02
0.025
10 30 50 70 90 110 130 150 170
Realized catch
Pro
ba
bil
ity
Degree of implementation succes = controllability
Aim: catch of 100
Inserting implementation uncertainty
State ofnature
Measure-ment / study
Realised effect
ActionSatis-faction
Action
1000 1500 2000
500 0.1Realised 1000 0.8 0.1effect 1500 0.1 0.8 0.1
2000 0.1 0.82500 0.1
Fisheries system: several optional control tools
ProfitPriorstock
CPUE
Fishing mortality
Demand
Natural mortality
Price
Value of the catch
Costs
Income
Catchability Production capacity
Other stressing factors
Incomefrom othersources
Posteriorstock
Environ-mentalfactors
Catches of other stocks
Fishingeffort
Yield
Fishing capacity
Number of fishermen
Predators
Equipment
Taxes
Value of perfect information: Perfect control
Mesh size
Prior variable 120 mm 140 mm
Recruitment process 11.7 5.8Growth 0.04 0.00Biomass criteria 0.00 0.00All variables 12.9 7.6
Bigger mesh size: system becomes more information robust
Doing has an effect on the need of knowing
Kuikka, 1994
Planning of management and monitoring by a meta-model
NaturalMortality
Catchability
Fishing mortality
YieldHerring
recruitment
Cod biomass
Water quality
Effort Cod fisheriesmanagement
Model 1
Model 2
Model 3
Which variables must be monitored, if I use variable A as a control variable ?
Some general conclusionsUsually:
1) The closer the control (decision variable) is to the objectivefunction, the better is the control
2) The closer the information link is to the essential source of uncertainty and the better is the controllability of the system, the higher is the value-of-information
3) The closer the monitored variable is to the objective, the easier it is to evaluate the success of your management
Part III: human aspects
Uncertainty Rowe (1994):
• Temporal uncertainty: future and past states• Structural uncertainty (uncertainty due to
complexity, related to control)• Metrical uncertainty (uncertainty in
measurements)• Translational uncertainty (uncertainty in
explaining uncertain results)
Implementation succes
Succes of management: dependent on fishermen
Identification of effective ”social impact tools”
Identification of sources of commitment
” Social capital” in the fishermen’s organisation
Is the complicated science needed only to convince/impresscolleagues: do we pay a high price on commitment side of actors?
What is good applied science ?
Management of humans
Control:rules, money, info
Knowledgeof individuals
Values and aims of individuals
Behaviour of individuals
Aims of society
Uncertainty of nature
Reaching of the aims
Number of recruits per one spawning fish in one year
Mean: 0,6 recruits per one spawning fish and year
0
0,5
1
1,5
2
0 100000 200000 300000 400000
kutukannan biomassa (t)
rek
ryyt
tejä
/ k
ute
va y
ks
ilö (
kp
l)
Impact of SSB on the number of recruits per one spawning fish and year in the Bothnian Sea herring stock
Peltomäki 2004
Recruitment size and maturity size & ”spawn at least once policy”
Maturity lengthDecrease of freq. of other managementactions
”Biological safetymargin ”
Recruitment size
Increase of freq. of other managementactions
% SPR and recruitment size: argumentation for fishermen
Some final points: logic of insurance systems and the message from
economic studies
Logic of insurance: pay to reduce uncertainty
0
5000
10000
15000
20000
25000
30000
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Fishing effort
Yiel
d/In
com
e
Economic view
0
2000
4000
6000
8000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Costs
Income (kg or kg * euro)
Profit
Spawning stock
Fishing effort