Supply Chain Design and
Planning
Lecture 1
In the supply chain management network, there are three challenges.
(1) coordination among partners to (2) match supply with demand (3)
under uncertainty and risk.
See the figure on the right. Fashion stores as SHEIN for example have a
fashion related demand, which is quite predictable. Therefore, their
supply chain has to be efficient.
SHEIN uses micro batching; they only have batches of 50/100 units of a
type of clothing. Like this, they can “test” whether the product is
popular and then will ask for more. They also keep supply chain
relationships. SHEIN also shares data with their suppliers, which
increases efficiencies.
Reshoring is bringing manufacturing operations in the country of origin of the company. But do you
want to have necessary products (like vaccines) in another country as your own? An example is Bath
& Body Works that produces soaps (basic product). They started the process of reshoring. They had
difficulties to convince the suppliers to relocate the production of raw materials near the production.
Reasons for the difficulties are dependencies and big investments. The company produces small
batches, which is more expensive. If you collocate your suppliers, you can get economies of scope. If
you want to produce small batches / be agile, then you need to reduce your fixed cost per shipment.
This can be done by reshoring. Economic order quantity=
√ 2∗¿ cost per shipment∗sales
per unit inventory holding cost
you want economies of cope, you need to reduce batch sizes, and then you need to downscale your
. So if
fixed cost per shipment (through reshoring).
Another topic is sustainability. For example, the stuff you return probably isn’t restocked and sent
back out to another hopeful owner. Why do companies allow us to return then? Because otherwise
you would not buy if you cannot return. Though, it is very costly.
Another topic is that airlines now provide an “all-you-can-fly” pass. They only offer if there are free
seats. They try to promote this pass very much. But does this not cannibalize the “normal” tickets?
You can argue that they only give empty seats. Though, the normal tickets are likely be sold less
(cannibalizing). Probably you’d spend more then. Moreover, the last-minute customer that would
normally buy a seat anyways on the last day/week, can now not buy anymore. This is a lot of lost
revenues. And how do you determine this price?
Then, there is AI. There are privacy issues and they can try to cheat the AI system.
Lecture 2 – preparations book & case
Sport Obermeyer is a fashion skiwear manufacturer. Wally needs to decide on the specific production
quantities for each skiwear item the company would offer in the coming year’s line. Market response
is crucial, but yet unknown for the last year’s items. He asked a committee of 6 professionals to
estimate next year’s popularity on their problems to estimate demand and manufacturing.
1
,He can allocate production between Hong Kong and China. Last year, parkas were made by a third in
China. This year, the planning is to produce half of its parkas in China. Labour costs are low there, but
there are concerns about quality and reliability of Chinese operations. Also, they require larger
minimum order quantities than those in Hong Kong, and they were subject to stringent quota
restrictions by the US government. In Hong Kong, 200 workers were higher for the factory’s first full
year of operation, mostly coming from the local community and distant towns nearby. Most were
new and were trained in the plant. Planning had been challenging here, since demand, worker skill
levels, and productivity levels were all difficult to predict.
Obermeyer developed into a preeminent competitor in the US skiwear market. Parkas were
considered the most critical design component of a collection. They segmented their customers in
gender and age and type. They competed by offering an excellent price/value relationship, targeting
the middle to high end of the skiwear market. Klaus (the manager) believes in providing value to
customers and emphasizing trust in people. Wally is the son of Klaus and relies more on data
gathering and analytical techniques, whereas Klaus takes a more intuitive style that was heavily
informed by his extensive industry expertise.
The clothes were mainly sold through ski-retail stores, located either in urban areas or ski areas.
Nearly two years of planning and production activity took place prior to the actual sale of products to
consumers. Europe is a good indicator of future American fashiosn, according to Klaus.
Supplier lead times are quite long (can be 90 days). Demand can be forecasted with great accuracy
(exhibit 5). The ship from Hong Kong warehouse to Seattle (from which they were trucked to the
distribution centre) takes 6 weeks.
When goods are popular in the demand season, sometimes Obermeyer could add the products from
their stocks to the stores. Unpopular items were sold at a discount. Items that had to go to the next
season, were sold at a loss.
The supply chain is:
Textile & accessories suppliers apparel manufacturers Obersport Sport Obermeyer retailers.
Obersport prepositions (purchases prior to the season and hold in inventor) fabric as part of a wider
effort to cope with manufacturing lead times. Later on, Obermeyer would spacify how it wanted the
fabric to be dyed and or printed. Though, Obermeyer had to take possession of all (uncoloured)
fabric. Workers in Hong Kong worked about 50% faster than their Chinese counterpartes, and the
workers could do a broader range of tasks. Units of parkas sold would earn 24% of the wholesale
price (pre-tax). If it could not be sold in that season, a loss of 8% would occur.
It occurred that if the buying committee had the highest level of agreement, the demand forecast
was most accurate.
This year, Obermeyer expected to produce about half of all its products in China. Longer term, Wally
wondered whether producing in China would constrain Obermeyer’s ability to manage production
and inventory risks. Would China’s larger minimum order sizes limit the company’s ability to increase
the range of products it offered or to manage inventory risk? Was Obermeyer’s trend toward
increased production in China too risky given the uncertainty in China’s trade relationship with the
United States.
Questions:
2
, 1. How should Wally think (both short-term and long-term) about sourcing in Hong-Kong versus
mainland China? What kind of sourcing policy do you recommend?
2. What operational changes would you recommend to Wally to improve performance?
3. Can you come-up with a measure of risk associated with your ordering policy? (this measure
should be quantifiable).
4. Using the sample data given in Table 2-20, make a recommendation for how many units of
each style Wally should make during the initial phase of production (assume all 10 styles are
made in Hong Kong and ignore price differences among styles). Wally’s initial commitment
must be at least 10,000 units.
IN CLASS CASE DISCUSSION
How should Wally decide on how much to order for each of the styles? He should look at:
- Trends from historical data (demand)
- Production sites of details (supply constraints)
- Look at product category of “Parkas” (most important)
- Know when to order.
He orders in November ’92 and places the first order (now). And he orders again in march ’94 (later
on). He needs to order before September ‘94 (to be in time).
So now, he needs to order the first time.
With regards to the suppliers, there is Hong Kong and China with differences in efficiency, volumes
and quality. Why do we already have to order now? He wanted to present the clothes to retailers
AND there is import quotas. But mainly, because of capacity constraints.
Parkas are the most standard and high-demand. He has asked to make a plan for production, which is
a RISKY choice. We can reduce risk by taking the standard/high-demand items, because less
information is needed. Here the risk is lower. This is the case for Parkas. The risk is that you produce
something that will not be sold (leftovers). Leftovers cost 8% of the price, because you need to
markdown. You reduce the price by 8%. Product characteristics related to this risks, price (more
expensive production costs, will cost more money (8% is then relatively more) and the demand
uncertainty of products (captured by the standard deviation), the average demand(producing 100
units if the average demand is 500, then this is less risky than producting 100 units for a product that
has an average demand of 100). These characteristics determine the riskiness of the products (in the
sense of risk of leftovers).
- How many units of each style should Wally produce?
In case of overstocking, there is an 8% loss (stated in text). Units of parkas sold would earn 24% of
the wholesale price. Thus costs of understocking is 24% of the price.
8% * price 24% * price DEMAND (given)
Style Price Over- Under- Mean StdDevn 2*StdDevn
stocking stocking
cost cost
Gail 110 8.8 26.4 1017 194 388
Isis 99 7.92 23.76 1042 323 646
Entice 80 6.4 19.2 1358 248 496
Assault 90 7.2 21.6 2525 340 680
Teri 123 9.84 29.52 1100 381 762
Electra 173 13.84 41.52 2150 403.5 807
3
, Stephanie 133 10.64 31.92 1113 524 1048
Seduced 73 5.84 17.52 4017 556.5 1113
Anita 93 7.44 22.32 3296 1047 2094
Daphne 148 11.84 35.52 2383 697 1394
Total 20001
In the text, it’s stated that the standard deviation of what is given times 2 is a good estimate for the
standard deviation of the real demand. That’s why we use the 2 x std.dev.
The optimal service level is 26.4/(26.4+8.8) = 75%. This is the same ratio for each item. ‘
To calculate the optimal order quantity you do
Norminv(75%;mean of the product; stddev x 2)
It’s the same as mean + z * demand std.dev z is then the critical value of the service level, using
the mean and std.dev.
Then you get the optimal order quantity.
In the text it is stated that in the first round, you only need to order 10,000 units. So we should not
order already more, because if we can postpone, our estimates of demand will be better. Thus, we
need to calculate how much we should order for each unit for the total to be 10,000 (in stead of
26360 if you calculate the optimal total quantity).
How do we do that? By: mean demand + std.dev * K. K is a number that’s the same for each style of
parkas. In this case, k = -1.06
DEMAND
Style Price Over- Under- Mean StdDevn x 2 Optimal Order First Order Qty.
stocking stocking Quantity (x)
cost cost
Gail 110 8.8 26.4 1017 388 1279 605
Isis 99 7.92 23.76 1042 646 1478 357
Entice 80 6.4 19.2 1358 496 1693 832
Assault 90 7.2 21.6 2525 680 2984 1804
Teri 123 9.84 29.52 1100 762 1614 292
Electra 173 13.84 41.52 2150 807 2694 1294
Stephanie 133 10.64 31.92 1113 1048 1820 1
Seduced 73 5.84 17.52 4017 1113 4768 2836
Anita 93 7.44 22.32 3296 2094 4708 1075
Daphne 148 11.84 35.52 2383 1394 3323 904
Total 20001 26360 10000
Notes Norminv VIA SOLVER (total
(75%;mean of of 10,000. Each
the product; unit is mean +
stddev x 2) std.dev * K)
Now, we can look at a feasible first order quantity. For some items, the minimum order quantity is
600. If we’d take that into account, we can order for each item that we’d order lower than 600 for
now order 600. Then you get the numbers of the 4 th column. You can also decide to only order for
items whose average demand is twice the minimum order quantity (MOQ) of 600. Then you only
take into account those and use solver again.
Style Optimal First Order Feasible First Alternative
Total Qty. (x) Order Qty. (x) First Order
4
