Supply Chain Frontiers issue #3. Read all articles in this issue.
As the National Football League gears up for a new season suppliers of NFL team replica jerseys are finding out how well their demand forecasting systems have scored. Driven by uncertainties such as team popularity and player trades, predicting which football jerseys will be in hot demand for the coming season is notoriously difficult. A research project looked at the problem and identified possible ways to avoid dropping the inventory planning ball.
The thesis project, completed this June, was carried out by John C. W. Parsons as part of an MIT Master in Engineering Logistics degree. As he discovered, consumer buying decisions are fickle and triggered by various reactions to players and teams and more general influences like the Christmas peak shopping season. Yet for suppliers there is scant room for error since buyers expect to find the right jersey when they decide to make a purchase.
In December 2000 the well known apparel company Reebok International Ltd signed a 10-year exclusive licensing contract with the NFL to manufacture, market and sell NFL licensed merchandise. "It is very important for Reebok to deliver products on time, particularly Reebok's ability to deliver hot-market items, a concern for retailers in all areas of the licensed business," observed Parsons.
Contract manufacturers make the jerseys and ship the items to Reebok's distribution center. The jerseys may be "dressed" with players names and numbers, or shipped as "blanks" without names and numbers which are added later in the company's Indianapolis, U.S. distribution center.
The Reebok purchasing department considers historical data, trade rumors and free agent status when setting targets for stock levels. However, the general approach is to account for demand at the team level and assign about 70% to 80% of the estimate to the top two or three players. "Approximately 25% of annual sales are known through early sales, or through forecasts received from large customers," Parsons said.
The buying cycle starts in July --14 months before the target NFL season begins. All of the jerseys ordered between July and September are blanks, with delivery planned for April. Items purchased from July to September are matched to retail orders. Between March and June purchases are made to position inventory at the Indianapolis distribution center in anticipation of retailer orders for the coming season. Said Parsons, given this complex demand pattern a major challenge for Reebok is how to "plan and manage inventory to manage costs while providing the flexibility required to meet demand for NFL replica jerseys."
Parsons, under the guidance of Professor Stephen C. Graves, used two modeling approaches to find out how this balance can best be achieved. The Simple Newsvendor Model considers demand for the selected players separately from the demand for blank jerseys. Each player is considered individually ignoring the option of using blanks to meet demand which is met by inventory for specific players. Blanks are only used for non-selected players. The second approach, Newsvendor Model with Risk Pooling, uses the Newsvendor method described above in combination with the risk-pooling opportunities afforded by including blanks as an option. In this model blank jerseys are stocked to fill demand for both selected and non-selected players.
The Newsvendor Model with Risk Pooling approach yielded the most favorable results for the planning of annual purchasing volumes. Total expected profit was eight percent higher than for the Simple Newsvendor model. There were inventory-related benefits as well, for example order quantities for dressed jerseys were significantly less with the addition of risk pooling.
In February, following retailers' initial order placements, enough information is available to generate a team and player level forecast. Using this forecast the Planning Manager can determine an optimal quantity for each player and team blank jersey, Parsons explained. Over the next several months purchasing programs can target these optimal quantities. As new information is available the forecast, and the model, must be updated. Player movements, increased early sales, or heightened expectations must all be incorporated into the forecast and managed to ensure that proper planning is carried out.
The analysis is not a cure-all but it does show how profits can be maximized by establishing the optimum balance between the use of dressed and blank jerseys. Neither is the model a substitute for industry expertise. As Parsons pointed out, "It cannot and should not entirely replace the experience, gut feeling, and art that every member of the Reebok team must have to understand and react in the professional sports business."
For further information on MLOG theses contact Becky Schneck, MLOG Admissions Coordinator, at bschneck@mit.edu.