html5-img
1 / 19

Sardonini L. 1 , Viaggi D. 1 and Raggi M. 2

Accounting for multiple impacts of the Common Agricultural Policies in rural areas: an analysis using a Bayesian networks approach. Sardonini L. 1 , Viaggi D. 1 and Raggi M. 2 1 Department of Agricultural Economics and Engineering, University of Bologna, Italy

vinny
Download Presentation

Sardonini L. 1 , Viaggi D. 1 and Raggi M. 2

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Accounting for multiple impacts of the Common Agricultural Policies in rural areas: an analysis using a Bayesian networks approach Sardonini L.1, Viaggi D.1 and Raggi M.2 1 Department of Agricultural Economics and Engineering, University of Bologna, Italy 2 Department of Statistics, University of Bologna, Italy 122nd European Association of Agricultural Economists Seminar Evidence-Based Agricultural and Rural Policy Making Methodological and Empirical Challenges of Policy Evaluation February 17th – 18th, 2011, Ancona (Italy) Centro Studi Sulle Politiche Economiche, Rurali e Ambientali associazioneAlessandroBartolastudi e ricerche di economia e di politica agraria Università Politecnica delle Marche

  2. Outline • Objective • Background • Methodology: Bayesian Networks (BNs) • Results from a farm/household survey in 9 EU countries • Discussion

  3. Objective • Discuss the potential use of Bayesian Networks to represent the multiple determinants and impacts of CAP in rural areas across Europe: • Analysis of stated intention to farming in 9 EU countries (micro level data)

  4. Background 1/2 Tools for evaluating effects of CAP are wide and heterogeneous: • high number of drivers • high number of potential dimensions (economic, social and environmental issues) • complex behaviour

  5. Background 2/2 Problems due to the complexity of relationships: • non-linear • too many variables • correlations among explanatory variables • multiple variables outcome • missing data

  6. Bayesian Networks (BNs) Some application fields: • Artificial Intelligence (first field): NASA, NOKIA • Sociology: Rhodes 2007 • Medical diagnoses: Kahn et al. 1997 • Environment: species conservation (Marcot et al. 2006), water (Zorrilla et al. 2010) • Land Use (Bacon et al. 2002)

  7. Bayesian Networks (BNs) • Simple and useful tools for modelling predictions and aiding resource managment decision making • Direct Acyclic Graph (DAG) where the nodes are random variables and the arcs represent direct connections between them (under conditional dependence assumptions)

  8. Bayesian Networks (BNs) • Example from Charniak 1996 Family-out Bowelproblem Input parent nodes Dog out Light on child node causal link Hearbark outcome child node

  9. BNs: advantages • Graphical construction interface • Incomplete database • Learn from data • Prior information • No linear relation • Could combine empirical data and expert judgement • Multiple outcomes

  10. BNs: methodology • Assuming a set explanatory variables pa(x) • Computation of P(xi|pa(x)) • Estimation using EM alghorithm: • Maximization of the log-likelihood • Iterative process • Update the posterior probability Bayes theorem

  11. Case study • Around 2000 farm-households interviews in 9 EU countries (telephone, face-to-face, postal) • European project CAP-IRE “Assessing the multiple Impacts of the Common Agricultural Policies (CAP) on Rural Economies”, 7th FP (SSH-216672) • Questions about farm and household (social characteristics, structural aspects and future intentions) • Policy scenarios: • CAP after 2013, No-CAP after 2013

  12. BNs: Application 1/2 • Variables used in the network: • Current farm/household characteristics • Multiple outcomes in terms of:

  13. BNs: Application 2/2

  14. Net description • The causal relationships derivebyWPsresults and economictheory • INTENTION is a key node • Currentcharacteristicsinfluence the INTENTION and all the outcomenodes

  15. BNs: Result (CPTs) Future stated plan to: • Adopt at least one INNOVATION_01: • young with a degree or old with high level of SFP and education • Increase the LAND_OWNED: • medium and medium-large farm size, rented-in already land and with at least two fulltime household members • Increase in MACHINERY: • increase in land and adopt at least one innovation • Increase in PESTICEDES: • livestock and mixed specialisation, SFP in the class 150-|500€ and increase the land • CHANGE_SELLOUTPUT • increase in land and adopt at least one innovation

  16. BNs: Results Effect of scenario (Cap/No-Cap) • Exit frequency increases in No-Cap (from 21% to 30.6%) • The adoption of at least one innovation decreases in No-Cap (from 28.9% to 25.5%) • The increasing in land size decreases in No-Cap (from 19.2% to 17.2%) • The increasing in the fulltime household decreases in No-Cap (from 19.35 to 18.1%)

  17. BNs: Accurancy • Error rates: percentage of missclassified between observed and predicted

  18. Discussion • Results • Coherence between the outcomes and the expectations • The older show a larger likelihood to quit farming activity • Good fit of the net in terms of low error rates • Further developments • Policy simulation: simulate the multiple outcomes from farming under different exogenous conditions

  19. Thank for the attentionlaura.sardonini@unibo.it

More Related