Canadian biofuels model

Comments

15 comments

  • Avatar
    Candi Clouse, PhD

    You are right to notice that high intermediate input rates will mean Type 1 multipliers, ergo higher indirect effects. Note that the “miscellaneous food manufacturing” sector in Canada has a high IE/Output rate, too, so that will compound the result of large indirect effects. Also, if TOPI and OPI make up a relatively small share of Value Added, that will create higher induced effects. In large study areas (such as multiple provinces in Canada), Output multipliers greater than 2 are common, so having indirect and induced effects that are larger than direct effects is not particularly concerning; that said, indirect effects that are orders of magnitude larger seem unlikely (albeit not impossible), even with high input rates.

    A couple questions:

    1. Are the spending patterns being adjusted for local purchase percentages (set either to a user-specified value or to SAM value=RPC) rather than to the default of 100%?
    2. Can you provide some more details about cases in which indirect effects are orders of magnitude higher?  What are the event inputs and what is the region (if not an aggregate of those specific provinces), and what are the results?
    0
    Comment actions Permalink
  • Avatar
    mrhaynes

    Thanks for your quick response. 

    Here are my answers to your questions:

    1. Yes, we adjusted the spending patterns to the SAM value.

    2. In the largest study area (all provinces combined) employment and labor income, in particular, had VERY high multipliers. I suspect that is because the biofuel producers that we received data from reported very low employment levels and wages relative to their operating costs. In some cases, wages/benefits represented only 2-4% of total output. I think I want to double check with the producers that those values are correct, but the result is that (in the largest study area) labor income had a multiplier of 20+. 

    The one I'm most concerned about is for the biodiesel production. This is biofuel that is produced from animal fats and vegetable oils. We attributed the feedstock costs to sector 23, but now I'm thinking that perhaps we should distribute those costs more broadly to some of the other related agricultural sectors (e.g. crop and animal production). I'm less familiar with the Canadian sectors, so I have been using a bridge table to translate U.S. sectors to the appropriate Canadian ones.

    Output multipliers seem more normal: in all study areas, they fall mostly above 2, but not more than 3.

     

     

    0
    Comment actions Permalink
  • Avatar
    mrhaynes

    Hi again,

    I just realized that one of the reasons my labor income multipliers are so out-of-whack is because I didn't include proprietor income in my direct effects, only employee compensation. However, we didn't specifically request estimates for proprietor income from producers, so we will have to estimate. I will use the social accounts/industry balance sheet estimates for some similar sectors.

    On the other hand, I also failed to model proprietor income as a separate activity, so that will increase indirect/induced/total effects. I assume the net result will be a decline in the multiplier overall? Time will tell...

    0
    Comment actions Permalink
  • Avatar
    Candi Clouse, PhD

    Although an output multiplier of 20+ certainly would raise eyebrows, labor income (and employment) multipliers of approximately 20 are also fairly common and tend to occur in industries with high IE/Output rates.

    Are you doing ABP with a Labor Income event? We just want to ensure that the labor portion isn't tied to an industry.

    0
    Comment actions Permalink
  • Avatar
    mrhaynes

    Yes, we are doing ABP with a labor income event. Glad to hear that the results are not highly unusual. I will make the adjustments I talked about (re: proprietor income) and otherwise, we'll continue to move forward with the methodology we have been using so far.

    Thanks for all of your help,
    Monica

    0
    Comment actions Permalink
  • Avatar
    mrhaynes

    Hello again,

    We have now completed our analysis and the multipliers ended up being a bit more reasonable, ranging from 5-9, depending on the biofuel type and the study area. However, the client still feels they are larger than they anticipated. Their comment was, “The employment multipliers are very large (larger than we have claimed in the past) and we suspect that this is generally related to the employment in feedstock production. This is difficult to justify because we are also reporting that there will be no expansion in cropland and that more feedstock can be produced through higher yields. The additional employment that is generated from higher yields is minimal, maybe some additional trucking or some crushing for veg oil.”

    I informed them of the intermediate expenditures and the cascading effect through multiple industries with high IE as well as the input of FTE vs IMPLAN’s ‘workers’ estimates.

    However, I was not able to fully address their point about meeting increased feedstock demand without cropland expansion and enabling higher yields which doesn’t correspond to many ‘new’ jobs. My question for you is this: is there a way to manually adjust the employment multipliers for the impacted industries in question, based on the information shared by the client?

    Thanks for your assistance.

    Monica

     

     

    0
    Comment actions Permalink
  • Avatar
    maria_lucas

    Hi Monica,

    In looking at the total employment impact and total output impact in the feedstock industries, an output per worker value can be calculated by dividing total output by total employment in the industry. If the Output per worker seems too low because you'd expect there will be higher yields, then you could assume a different Output per worker, and recalculate Employment impacts based on Output divided by your assumed Output per worker. We recommend noting such assumptions and edits to the results. 

    With expansion of cropland being restricted, an alternative to higher yields in the region would be more imports of the feedstock. In this case, you could reduce the effects in the feedstock industries evenly across all indicators (Employment, Labor Income, Value Added, Output) by some percentage. If there is a maximum Output or Employment you believe to be feasible, I'd would suggest reducing effects to the Industry (proportionally across indicators) to match the expected maximum supply from these industries within the region.

    A question I have after reading through your approach so far is, are you using an Industry Spending Pattern Event Type? and do the Sum of Percentages in your Spending Pattern equal 100%? 

    Thank you,
    Maria 

    0
    Comment actions Permalink
  • Avatar
    mrhaynes

    Thank you for the guidance. I will talk to the client and see which method you've suggested seems most appropriate and go from there.

    In response to your question... yes we are using an industry spending pattern event. The sum of the percentages in the spending pattern do not equal 100% - they exclude wages and other value added elements. The industry still has a very high percentage of intermediate expenditures, though. I believe the total is higher than 90%.

    0
    Comment actions Permalink
  • Avatar
    maria_lucas

    Sounds great, Monica. Let us know if we can help answer any other questions you run into.

    Happy Holidays,
    Maria 

    0
    Comment actions Permalink
  • Avatar
    mrhaynes

    Sorry, one more question:

    Regarding your suggestion to "assume a different Output per worker, and recalculate Employment impacts based on Output divided by your assumed Output per worker". - is that something we would do manually in the results themselves? Or is it something we do to the model?

    Similarly, when you suggest we "reduce the effects in the feedstock industries evenly across all indicators (Employment, Labor Income, Value Added, Output) by some percentage" - would that be a reduction in coefficients for the sectors we've identified as being feedstock purchases?

    Thanks again,
    Monica

    0
    Comment actions Permalink
  • Avatar
    maria_lucas

    Hi Monica,

    In the first approach, edits would need to be made manually to the results to assume a different output per worker. 

    As for the second suggestion of "reducing the effects", I was suggesting a manual edit to the results, but you could make an adjustment in the impact setup. Instead of reducing the coefficients for the sectors, you could alternatively reduce the Local Purchase by switching from "SAM Model Value" to "User LPC" and entering a value smaller than the SAM value. Enter the value that appropriately reflects the portion of these commodities that will be sourced from the region.

    Thank you,
    Maria 


    0
    Comment actions Permalink
  • Avatar
    mrhaynes

    Hello Maria,

    Thanks for your response. I just want to make sure I get this right...

    The client doesn't seem to think the increased crops will be imported. So, I plan to incorporate the first approach you suggested, and manually edit the results so that we are assuming a higher output per worker in the crop and animal production sector. I plan to reduce the employment values shown in our detailed indirect effects for the "crop and animal production" sector by some amount, so that the new indirect effects in that sector reflect a higher output per worker. I haven't yet determined what that new ratio should be, but I will work with the client to determine it. I just want to make sure that I'm using the right method. Should I make the change to that one particular sector only, within the detailed indirect effects, and then recalculate the totals accordingly? Or am I misunderstanding?

    Thanks,
    Monica

    0
    Comment actions Permalink
  • Avatar
    maria_lucas

    Hi Monica,

    Yes, for this approach you should make the change to Employment in that one particular sector and recalculate the total Employment impact. Reading back to a comment you placed emphasis on: "we are also reporting that there will be no expansion in cropland" , it may be reasonable to replace the total detailed Employment impact on the sector with the total Employment in the sector and region. You could use the total Employment in the sector from the Model Overview, but the preferably the Employment level in the sector at the time of the activities being analyzed would be best. 
     
    If you'd like to assume a new ratio of Output per worker and apply this to the results, then the existing Employment value(s) in the particular sector should be replaced with quotient of the Output (in the sector) divided by the determined new ratio of Output per worker for the sector. With the Employment figures updated in the particular sector, re-sum for a new total Employment. I'd recommend applying the change to all effects to the industry, which I'd expect are largely in the indirect effects. The existing assumed Output per worker can be calculated as the total detailed Output in the sector divided by the total detailed Employment in the sector. 

    For example, let's say the total detailed Employment in the sector is 10 and the total detailed Output is $1M, then the assumed Output per worker ratio is $100K. If I'd like to assume a new Output per worker ratio of $200K per worker, then I'd replace the Employment value of 10 with $1M/$200K = 5. 

    Keep in mind if no change is made to Labor Income Effects in the sector the effective Labor Income per Worker ratio in the sector will also be affected. 

    Thank you,
    Maria 

     

    0
    Comment actions Permalink
  • Avatar
    mrhaynes

    Thank you for this detailed response.

    I have another question. As I look at the output per worker for the industry, there are basically two values. the first, taken from the study area data, is $222,219 for crop and animal production. However, when I calculate the output per worker based on the indirect effects in our results, that value is higher: $255,558. I assume that difference is due to the multiple rounds of spending that are happening in the model, but I just want to confirm. If I am adjusting the results based on output per worker, I assume I am adjusting the larger value ($255,558), correct?

    0
    Comment actions Permalink
  • Avatar
    maria_lucas

    Hello,

    You are correct. I expect that difference is due to the difference in Dollar Years. The study area data is in the Dollar Year of the Data Year. Your results are shown in the Dollar Year specified on the results screen. Output gets inflated/deflated based on Dollar Year, but Employment does not, so the ratio of Output per worker is expected to change when viewing in a different Dollar Year than the Data Year. 

    (correction in explanation made after original response posted)

    Maria 

    0
    Comment actions Permalink

Please sign in to leave a comment.