Models of the forest products sector simulate production, consumption and trade of forest products in a recursive market equilibrium. Such models are often seen as "black boxes" by non modellers. What are the different ways to validate their input data, methodologies, and output scenarios?
Some authors suggest (because of reflexivity relationships) that in the real-world there are not market equilibriums [http://en.wikipedia.org/wiki/Reflexivity_%28social_theory%29]
Testing models is always a concern especially when you are working with a long term product like forests. What i like to do is after making sure that the model runs and provides an answer that can be defended is to put the model through a series of what ifs. Ask yourself is the model responding as one would expect and if not, why not. In agriculture, running an analysis in the 70's, an unrestricted model run indicated that soybeans would be a huge crop in areas that soybeans were not traditionally growing. In this case, as it turns out the model was correct and expectations were not.
Thank you both for your answers,
@Ricardo: reflexivity is an interesting new concept for me.
@Burton: put the model "through a series of what-ifs" sounds good.
I am running an existing forest sector model based on FAO data. I calibrated it on 1992. At the begining of the study, I'd like to list a range of testing methodology, from an "expert assesment" of the input data, to model validation with sensitivity analysis. I'm looking for literature in the field of testing of validation of models for resource economics.
I further advice you to watch this talk from Nassim Taleb:
http://www.nsf.gov/news/mmg/mmg_disp.cfm?med_id=73816&from=mmg
Notice mainly what he says about stress tests and others alike.
I agree with Burton and Ricardo that stress test is a necessary. This, as always, entails applying 'common sense', assessing wherther the modelling output are internally consistent as well as inline with the best knowledge of the workings of the real world. In any case, one should, as much as possible, be prepared to question and discuss the modelling results. If possible, comparisons with other forest sector models is also useful, thouh most of them are very similar in their structure.
A common short-coming of the existing forest sector models is that they only cover the 'traditional' wood-based products (though also energy is now being included). 'New products' are very hard to include, due to very limited knowledge as regards demand, mainly. These products needs to be modelled in a qualitative way
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress