Shipping the Chickens

I've Blog posted on our draft plan for the removal of Supply Management from Canada.

That plan allows everybody to agree on the farm-gate price to be paid for chicken (or the other SM commodities of  eggs, chicks, turkeys, & milk).  However, there is a geographical separation between the producers and the processors.  Somehow, we have to arrange for the transportation between the two parties.

The truncated k-double auction can match up a specific producer with a specific processor, but based on the randomness of the auction process and the resulting bid-ask matching, that matching could cause unrealistic shipping of chicken everywhere across the province.  It is therefore best for the auction process to be supplemented by a least-cost shipping schedule.

How do we do that transportation, and who pays for it?  Obviously, it is better to minimize shipping costs (sorry truckers).

Ontario's Quota and Non-Quota Chicken Growers

 We have to optimally hook up all of the red dots above (chicken farmers), with all the red dots below (chicken processors):

Federally-inspected Meat Processing Plants

Plus the following Provincially-inspected meat processing plants:

Provincially inspected meat processing plants

Once a producer or processor are geographically located, it cannot be easily moved, yet things change.  Populations shift, vehicle fuel prices go up & down, producers & processors get into disputes that force them to deal with others than the geographically closest match, and so on.  If somebody decides to grow chicken in Thunder Bay, how do we (should we) be shipping those chickens to a processing plant in Toronto?

What about inter-provincial movements?

We have a few options:

1. We do separate auctions for each region.  This complicates the auction process greatly.  It may cause the balkanization of the provincial chicken market (ie. sub-division, non-cooperation, competition, abandonment).  Some auctions may get shunned by important producers or processors.  There may be large price differentials between the different auctions, putting one side at a significant competitive dis-advantage.  Popular auctions are over-loaded while unpopular auctions are under-serviced.
   
   
2. We leave those transportation details for each producer-processor pair to work out for themselves.  The auction sets the price for the chicken, transportation extra.  This might work most of the time, but it will probably be like musical chairs when the music stops; a mad scramble to grab the nearest chair, somebody gets hurt, or you've left standing and the nearest empty chair is at the far end of the room.  Not nice when you have to do this every quota period, and your life depends upon you making it there before somebody else.
   
   
3. We have FOB shipping hubs defined all across the Province.  The producer includes the cost of getting his product to the nearest FOB shipping hub in their auction bid, the processor pays to transport from their nearest FOB shipping hub to their factory and includes that price in their auction bid, then the bookkeeping cost of inter-hub transfers is pooled and shared equally by all.  I wouldn't want to be the one who decides how many and where these FOB shipping hubs will be located, for nobody will be happy with the decisions made.
   
   
4. The shipping costs are pooled for the entire province.
    
5. The shipping costs are pooled for each of the regions within the province.

Dijkstra's Algorithm is the standard algorithm that is used to find the shortest path between two points (eg. GPS mapping and driving directions).  When we have multiple sources and multiple destinations, it's a little more complex.  Good thing IBM has already solved that problem, way back in 1996.  They used a modified A* heuristic algorithm to estimate the Dijkstra optimum path in 30% less computational time. There is also the Floyd-Warshall Algorithm. Neither of these is exactly what we will need to solve the Multiple Pairs Shortest Path Algorithm ("MPSPA") that we will require for provincially-optimum chicken shipments, so some computer software development work will have to be done on this issue.

Once the auction results are defined, the winners' names & locations will be entered into this MPSPA shipping optimization software, and it will calculate the province-wide optimum pairing up of buyers & sellers; defining which supplier gets to ship to which processor.  Each of the parties are then notified of who bought & sold for each bid-ask pair.  If there are specific classes of chicken sold, then multiple runs of MPSPA, one for of the various classes of chicken, will have to be calculated.

While this gets us a province-wide optimum shipping schedule, the cost of shipment is still significant, as well as the scheduling of the catching crews.

Do we need to have all producers and all processors treated the same, no matter where they are located in the province, removing geography from the equation?

For example, let's say a producer is right next door to a chicken processing plant, and these two love each other, and have been supplier-customer forever.  Perhaps, for the good of the overall optimum on a provincial basis, it is better to break up these kindred spirits, and they will be required to buy and sell to somebody else who is a little farther away, and somewhat of a stranger.  This will cause some frustrations & higher costs for both of the original pair.  If we wish to get a provincial minimum cost, shouldn't the minimum cost be pooled across the province, so these two chicken lovebird don't personally pay more for the good of everybody in the province?

I therefore recommend that shipping costs be minimized on a provincial basis using MPSPA, then the shipping costs are pooled across all of the province, and added as a $/kg cost to all chicken transactions.  This makes all farms and all processors on an equal footing cost-wise, no matter where they are located in the province.