Water, Land & Ecosystems Intervention Decisions
Where are the high information values?
Keith Shepherd & Doug Hubbard
Nov 2012
Vagen
Water, Land & Ecosystems (WLE)
Organizing research around a conceptual framework of basins
and landscapes
WLE Decisions
Our strategic objectives = our system level outcomes: (i) decrease food insecurity(ii) manage environmental resources(iii) reduce poverty among farmers(iv) increase nutrition, health and wellbeing We aim to improve stakeholder decisions on policies, intervention programmes and intervention designs through research What information has high value for improving decisions to achieve these outcomes?
•How to prioritize research under uncertainty
•Which interventions will reduce risk, increase security, and improve lives the most? What are the trade-offs between competing objectives, like agricultural productivity and the environment? What are the risks of intervention failure?How to measure and monitor development outcomes
•Potentially huge investments in monitoring but not all metrics will be of equal value to support intervention decisions. How should we determine what data gathering costs are justified?How to show the value of research
How can we show how the expense of research is justified by better intervention decisions and improved outcomes?
Challenges Facing Researchers
Development of systems to measure the impact of CGIAR investments (of relevance to DFID as a significant funder) at the level of the 4 system outcomes. Mechanisms to analyse the impacts and trade-offs associated with sustainable intensification at different scales (sub-national, national, regional).Value for money metrics for measuring agriculture, ecosystem and poverty and nutritional outcomes.
Interests of donors
Why must quantify uncertainty
•Averages are wrong on average •Uncertain events (floods, droughts, erosion, market fluctuations) •Security is a development outcome (food/nutritional security; risk is the complement of security) •Value of information
Walsh
How much information do we need?
What defines whether information is unreasonably expensive? What is the value of doing one more survey or experiment, or creating another database? Organizations often spend 10 times the value of information on surveys and trials, etc
[Ron Howard]
We need a method to quantify information value
How to make preferences explicit
Objective trade-offs •The trade-offs between productivity, ecosystem and welfare outcomes Valuation of outcomes (Preferences, Policy) •Valuing one outcome relative to another (production vs environment) •Time (benefit now versus later) •Uncertainty (risk aversion) •Equity (increasing income of poor worth more than non-poor)
Making preferences explicit improves transparency and multi-stakeholder decision processes
Applied Information Economics
Hubbard Hubbard
© Hubbard Decision Research, 2012
Uses of Applied Information Economics
AIE was applied initially to IT business cases. But over the last 17 years it has
also been applied to other decision analysis problems in all areas of Business
Cases, Performance Metrics, Risk Analysis, and Portfolio Prioritization.
• Prioritizing IT portfolios
• Risk of software
development
• Value of better information
• Value of better security
• Risk of obsolescence and
optimal technology upgrades
• Value of infrastructure
• Performance metrics for the
business value of
applications
IT
• Risks of major engineering
projects
• Risk of mine flooding
Engineering
• Movie / film project selection
• New product development
• Pharmaceuticals
• Medical devices
• Publishing
Business Civilian Government
• Environmental policy
• Procurement / auction
methods
• Grants management
Military
• Forecasting battlefield fuel
consumption
• Effectiveness of combat
training to reduce roadside
bomb / IED casualties
• R&D portfolios
Payback is 20:1 to 300:1
The AIE Process
Identify important metrics for monitoring implementation
Improve the intervention design to reduce chance of negative outcomes
Hubbard
Forecasting intervention impacts
Value of information
Game theory provided a formula for the economic value of information over 60 years ago: Expected Opportunity Loss = the chance of being wrong x the cost of being wrong Expected Value of Information is the reduction in the EOL as a result of the additional information.
AIE Empirical Evidence • We are not as clear as we think on the decisions we are trying to
influence
• Expressing uncertainty dissolves assumptions & allows all benefits,
costs and risks to be included, however intangible (especially
environment!)
• We need calibrating to reliably estimate probability distributions
• There are usually only a few variables with high information value
• We are often measuring the variables that have least economic value
• And completely missing the ones that do have value (e.g. tend to
measure costs but ignore benefits, which are typically uncertain).
• Measurement is uncertainty reduction, not a gold standard
• Often need different data than we think
• Often need less data than we think
• Even small reductions in uncertainty can have considerable value
© Hubbard Decision Research, 2012
Safe Drinking Water Information System
• The EPA needed to compute the ROI of the Safe Drinking Water Information System (SDWIS)
• As with any AIE project, we built a spreadsheet model that connected the expected effects of the system to relevant impacts – in this case public health and its economic value
Input sheet
Need for calibration training
Cash flow page
Risk report page
Cost-effective Measurement
• Fermi decomposition
Estimate no. of piano tuners in Chicago
= No. households (population/people per
household)
x % of households with tuned pianos
x tuning frequency per year / (tunings per day x
work days per year
• Secondary research - measured before? Historic
data
• Observation - sampling, tracers, experiment
[From Hubbard 2010]
Value of Information
A Probability Management System Decision modelling defines the metrics
Smart data - Smart decisions
Quantifying WLE Intervention Outcomes
Time value preference
Years
Environmentally rational
Poor farmer
Next steps Phase 1
•Analysis of 4 - 6 WLE intervention categories/cases in 2013
•The outcomes define the agro-ecosystem metrics databases
•Decision Analyst
Phase 2
•Standardized databases and stochastic libraries
•Generic intervention screening model (triage method)
Phase 3
•Develop intervention decision modelling platform (linked stochastic libraries, visualization tools)
•Analysis of WLE or CGIAR project portfolio
Smart data - Smart decisions