|
Module SCM103: Supply Chain Strategies II: Improving Responsiveness & Advanced Topics
|
|
|
| List Price: |
Duration: |
Word Count: |
Rating: |
Datasheet: |
| $60 | Approx.
1 hour | Approx.
8,400 words |  
(4.1 / 5) |  Download |
| Certificate: |
Counts toward Fundamentals of Supply Chain Management |
|
| Promotions: | Save $60 on "Fundamentals of Supply Chain Management" bundle; buy modules SCM102, SCM103, SCM104, SCM105, and SCM106 together and save $60 |
|
|
|
|
|
Improving Responsive Supply Chains: Accurate Response & Risk-Based Production Planning
Companies can improve responsiveness by adapting to fluctuating or uncertain demand. Two ways of doing this are accurate response and risk-based production planning.
The term accurate response refers to a series of concepts popularized by Prof. Marshall Fisher and his colleagues in their studies dealing with the apparel industry, involving a methodology for reducing demand uncertainty.1 One component of accurate response is noticing that for fashion items like skiwear, it is difficult to forecast which styles will be popular at the beginning of the season, but once the retail selling season starts, early sales are excellent forecasters of subsequent sales. The solution is to obtain such information early enough to be useful. One idea, used by Sport Obermeyer, was to create an "Early Write" program and invite selected retailers to Aspen, Colorado. Orders written during the Early Write program received an additional discount, but the information obtained by those orders was invaluable in determining which styles were likely to be the "hot" styles for that selling season.
A second component of accurate response is lead time reduction. We saw one way to do this in the semiconductor industry in our SCM102 module, by reserving capacity rather than specific SKUs; Sport Obermeyer's solution was to use international express mail services to send product design information to manufacturing partners in Asia. The additional cost of express mailing was more than justified by the competitive advantage they gained by saving just a few days' lead time (nowadays we'd use the Internet to do this, at very low cost). If you think the express mail solution is obvious, put yourself in the place of an office manager who is criticized by your boss for spending $1,000 on international express mail services last year! Unless you can point to a specific, tangible benefit, there will always be people in organizations who focus on the wrong measure, or a single narrow measure instead of a broader measure like profit or return on investment.2
A third component of accurate response is more complex; it involves estimating your company's forecasting accuracy about similar products. This strategy applies not only to the apparel industry; this analysis can prove useful any time you face highly uncertain demand subject to consumer tastes and/or changes in technologies and feature sets. Suppose there are six "experts" in your company who each create an independent forecast of sales for some similar products. For example, consider the following six sales forecasts for three new apparel styles:
|
| |
Expert Forecasts of Sales |
| Style |
1 |
2 |
3 |
4 |
5 |
6 |
| A001 |
10,000 |
8,000 |
9,000 |
11,000 |
12,000 |
10,000 |
| B002 |
4,000 |
4,500 |
3,500 |
4,500 |
5,000 |
5,500 |
| C003 |
2,000 |
5,000 |
1,000 |
8,000 |
3,000 |
5,000 |
|
|
You can assess their relative agreement among each other as a measure of how much your company as a whole knows about the popularity of a given product or style. If the six experts have sales forecasts that are, for example, within 20% of each other, we would say that style seems reasonably predictable. But if the six expert opinions differ by a factor of two or three (e.g., 1000 vs. 2000 or 3000), then we would say that style is unpredictable.
What can we tell about predictability of each style using the forecast data above? First, we'll determine the average, standard deviation, and scaled standard deviation for each style:
|
| |
Expert Forecasts of Sales |
Calculations |
| Style |
1 |
2 |
3 |
4 |
5 |
6 |
Average |
Standard
Deviation |
Scaled
Standard
Deviation |
| A001 |
10,000 |
8,000 |
9,000 |
11,000 |
12,000 |
10,000 |
10,000 |
1414.2 |
0.141 |
| B002 |
4,000 |
4,500 |
3,500 |
4,500 |
5,000 |
5,500 |
4,500 |
707.1 |
0.157 |
| C003 |
2,000 |
5,000 |
1,000 |
8,000 |
3,000 |
5,000 |
4,000 |
2529.8 |
0.632 |
|
|
If you use the standard deviation as a measure of predictability, the least predictable style is C003, and the most predictable style is B002. But we should adjust these values for the basic forecast levels for each style, since they differ by a large amount. The last column above, the scaled standard deviation, simply adjusts the standard deviation by dividing it by the average forecast for each style.3 Based on these scaled, or adjusted, values, we see that Style A001 actually is the most predictable style (by a small margin over Style B002), while Style C003 is still by far the least-predictable style.
How is this information helpful? You should be willing to produce the more predictable style early on, before the selling season, using what is called "speculative capacity" of the supply chain. Then, after the wholesale selling season starts and retailer orders start coming in, demand for the less-predictable styles becomes clearer, and ideally you have reserved some "reactive capacity" in your supply chain to manufacture whichever of those styles have the highest demand for this season. This is called Risk-Based Production Planning. Risk-based production planning, in this case, tells us to use early production capacity ("speculative capacity") for Styles A001 and B002, and reserve later production capacity ("reactive capacity") for Style C003 as demand becomes more apparent:
|
|
|
The concepts of accurate response and risk-based production planning are powerful tools when applied to innovative products. When these concepts were applied to an apparel manufacturer, markdowns and stockouts were reduced significantly, with a considerable improvement in profitability.4
1 "Making Supply Meet Demand in an Uncertain World," by Marshall L. Fisher, Janice H. Hammond, Walter R. Obermeyer, and Ananth Raman, Harvard Business Review, May-June 1994.
2 "SCM105: Performance Measures for Supply Chain Management", explores issues such as this in more detail.
3 The standard deviation is a measure of variability. The scaled standard deviation is a measure of relative variability; scaling the standard deviation eliminates the magnifying effect of overall forecast size on the standard deviation, allowing us to compare variability between items with significantly different forecasts. See Hausman, W. H., and Bierman, H. Jr., Quantitative Analysis for Management, Ninth Edition, McGraw-Hill, 1997.
4 Fisher et al, 1994, op. cit.
|
|
|
