Supply Chain Ll Bean
AUGUSTIN BOLZE March 16th March 2015
SCM – Dimitrios Andritsos – 9 :40
L.L. BEAN, Inc CASE
1. How does L.L. Bean use past demand data and a specific item forecast to decide how many units of that item to stock?
L.L. proceeds step by step to decide how many units of an item they stock. After negotiations, discussions, they obtain a specific item forecast, a “frozen” forecast. However they are not going to use past demand data on this precised item to know how much to stock of this item (moreover there are new items), but they are going to use the past demand data of all the items. What they do is that they compute the ratio (Actual demand / forecasted demand) for every item of the previous year. Then they get a range of past forecast errors as well as the frequency of each error. Then, they use this frequency distribution of past forecast errors as a probability distribution for the future forecast errors.
Once we have this distribution of forecast errors, they look at how much the profit of a sold item is against the lost of this same item at a liquidation price. The more the profit is big compared to the loss, the higher percentile of the previous distribution they are going to use to compute the frozen forecast to an actual commitment, stock. For instance, if an item is sold 40$ and costs 20$, the profit is 20$. Its liquidation price is we say 5$. Then the profit is 4 times bigger than the loss. Therefore, we are going to use a percentile above 0.8. It may correspond to a ratio of 1.4. if the frozen forecast was 2000 after negotiations, L.L. Bean is going to order to its vendors: 2000*1.4=2800 items.
Therefore we use past data from all of the items, no matter what they are (shoes, shirts…) to value the future stock of one precise item. We can understand the concerns of Rol Fessenden.
2. What item costs and revenues are relevant to the decision of how many units of that item to stock?
3. What information should Scott Sklar have available to help him
SCM – Dimitrios Andritsos – 9 :40
L.L. BEAN, Inc CASE
1. How does L.L. Bean use past demand data and a specific item forecast to decide how many units of that item to stock?
L.L. proceeds step by step to decide how many units of an item they stock. After negotiations, discussions, they obtain a specific item forecast, a “frozen” forecast. However they are not going to use past demand data on this precised item to know how much to stock of this item (moreover there are new items), but they are going to use the past demand data of all the items. What they do is that they compute the ratio (Actual demand / forecasted demand) for every item of the previous year. Then they get a range of past forecast errors as well as the frequency of each error. Then, they use this frequency distribution of past forecast errors as a probability distribution for the future forecast errors.
Once we have this distribution of forecast errors, they look at how much the profit of a sold item is against the lost of this same item at a liquidation price. The more the profit is big compared to the loss, the higher percentile of the previous distribution they are going to use to compute the frozen forecast to an actual commitment, stock. For instance, if an item is sold 40$ and costs 20$, the profit is 20$. Its liquidation price is we say 5$. Then the profit is 4 times bigger than the loss. Therefore, we are going to use a percentile above 0.8. It may correspond to a ratio of 1.4. if the frozen forecast was 2000 after negotiations, L.L. Bean is going to order to its vendors: 2000*1.4=2800 items.
Therefore we use past data from all of the items, no matter what they are (shoes, shirts…) to value the future stock of one precise item. We can understand the concerns of Rol Fessenden.
2. What item costs and revenues are relevant to the decision of how many units of that item to stock?
3. What information should Scott Sklar have available to help him