Holistic risk-return analysis for improving planning and performance measurement of agroforestry interventions

Abstract

A quantitative risk-based framework is presented for improving planning and performance management of agroforestry interventions at farm and project levels. The framework first identifies the goals of the various stakeholders involved and the intervention options being considered. Then the various on-site and off-site costs, benefits and risks associated with the intervention options are identified, including biophysical, social and economic factors. A combination of available data and expert knowledge is used to estimate probability distributions on all variables. Stakeholder's risk preferences are also quantified. Outcomes are presented in monetised form (e.g. distribution of net present value) providing insights into trade-offs among objectives and stakeholder groups. Partial Least Squares regression is used to identify which variables contribute to positive and negative outcomes, providing opportunities for improvements in intervention design. Value-of-information analysis points to areas where gathering further information would help clarify the decision, and provides a guide on how much it is worth spending to collect it. The analysis also points to which variables should be closely monitored during implementation. The framework is implemented using either Monte Carlo (MC) simulation or Bayesian Networks. An MC package in R software that generates the above analyses is already available. The framework is illustrated with examples at farm, project and policy level. Diagram of framework for evaluating agroforestry interventions

Full text

464
Agroforestry 2019
- Poster L12 Economics of AF
4th World Congress on Agroforestry
Strengthening links between science, society and policy
20-22 May 2019
Le Corum, Montpellier, France
Book of Abstracts
L12.P.20
Holistic risk-return analysis for improving planning and performance
measurement of agroforestry interventions
Shepherd K.1 (k.shepherd@cgiar.org), Luedeling E.2, Whitney C.2, Muchiri C.1, Tamba Y.1
1 Land Health Decisions, World Agroforestry Centre (ICRAF), Nairobi, Kenya; 2 Department of Horticultural
Sciences, University of Bonn, Bonn, Germany
A quantitative risk-based framework is presented for improving planning and performance
management of agroforestry interventions at farm and project levels. The framework first
identifies the goals of the various stakeholders involved and the intervention options being
considered. Then the various on-site and off-site costs, benefits and risks associated with the
intervention options are identified, including biophysical, social and economic factors. A com-
bination of available data and expert knowledge is used to estimate probability distributions
on all variables. Stakeholder’s risk preferences are also quantified. Outcomes are presented
in monetised form (e.g. distribution of net present value) providing insights into trade-offs
among objectives and stakeholder groups. Partial Least Squares regression is used to iden-
tify which variables contribute to positive and negative outcomes, providing opportunities
for improvements in intervention design. Value-of-information analysis points to areas where
gathering further information would help clarify the decision, and provides a guide on how
much it is worth spending to collect it. The analysis also points to which variables should be
closely monitored during implementation. The framework is implemented using either Monte
Carlo (MC) simulation or Bayesian Networks. An MC package in R software that generates
the above analyses is already available. The framework is illustrated with examples at farm,
project and policy level.
Diagram of framework for evaluating agroforestry interventions
Keywords: Decision analysis, Multiple objectives, Risk analysis, Participatory, Assessment.