I am working with the 2012 Canadian dataset. The primary study area I am using includes the provinces of AB, BC, MB, ON, QB, and SK. The study I am conducting is measuring the economic impacts of biofuel production on the provinces. I have production levels and estimated revenue for four types of biofuels. I created a unique industry spending pattern for each type of biofuel, based on feedback from producers on their estimated annual spending.
My question is, after running the models, the results are showing indirect and induced effects that are much higher than the direct effects. In some cases, the indirect effects are magnitudes larger. That is so unlike what I expected that I feel something must be wrong. However, I've double checked all of my inputs and they seem correct.
A couple things that I am suspecting might be behind the large results: First, the producers reported very low profit margins and low employee wages/benefits, so a big majority of their revenue (90%+) is spent on inputs. Second, of their inputs, a huge portion (between 65-80%) goes to feedstock (e.g. corn, wheat, fats, oils). For example, in one model, more than 75% of the industry's annual revenue is spent on feedstock (according to feedback from producers). However, because there is so little detail in the Canadian sectors, it's very hard to place those costs precisely in a sector that makes sense. In one example, all feedstock purchases are put into only one sector: sector 23 "miscellaneous food manufacturing". That is creating some strange results.
Back to my question... do the results I'm describing sound incorrect? Do you have any recommendations for how I might adjust the models to "smooth" out the ISP? One thought I had was to try to anticipate the second round of spending on feedstock purchases, but I'm not sure how exactly I would go about doing that.
Any guidance would be appreciated.
Please sign in to leave a comment.