BUSI 615 Week 2 Discussion Board 1 replies Example

Discussion Board 1: Replies to Classmates

BUSI 615 Logistics, Discussion Board 1 (two replies)

Thread by [Student Name]

The two classmates and their threads are composites standing in for the forum's own posts; the research cited is real, and the requirements follow Liberty's public course guide for BUSI 615.

Reply to Marcus

Marcus, your thread described an auto parts supplier where schedulers routinely change the statistical forecast before it reaches the master production schedule, and you argued that these overrides are the main reason the plant's MRP keeps releasing orders that are later canceled. That is a strong example, and this week's forecasting chapter gives us a way to test it. I think the evidence supports half of your argument: overrides do cause harm, but only certain kinds.

Fildes et al. (2009) examined more than 60,000 forecasts at four supply-chain companies, comparing the statistical forecast with the version after managers adjusted it. Adjustments were very common. The large ones, usually made because a manager knew about a promotion or a lost customer, tended to improve accuracy. Small adjustments did little good and often made the forecast worse, as if planners were tinkering to show involvement. The study also found an asymmetry: downward adjustments were more likely to help than upward ones, which the authors linked to optimism among people close to sales.

If your plant follows the same pattern, the schedulers' habit of nudging every number up by a few percent would explain exactly what you saw. A small upward change to the forecast for a part family becomes a larger master schedule, MRP explodes it into purchase orders for every component, and when actual demand comes in lower, those orders are canceled or pushed out. The cost of a small optimistic adjustment is paid many times over, once at every level of the bill of materials. The planners feel they have added judgment, but the system has amplified a bias.

That suggests a rule rather than a ban. The plant could allow overrides only above a threshold, say a change of 15 percent or more, and require a stated reason for each, for instance a customer launching a new program, then track the accuracy of adjusted and unadjusted forecasts each month. Within a quarter, the data would show whether the schedulers' knowledge adds value. Our chapter on demand management makes a similar point when it treats the forecast as something to be managed, with assumptions recorded, rather than simply produced.

The rule also has a cost the plant should weigh honestly. Recording reasons and reviewing accuracy takes planner time every month, and some schedulers will resent what looks like an audit of their judgment. Framing the review as a way to find out which of their insights the statistical model is missing, and then feeding those insights back into the model, may make it easier to accept. Over time the threshold itself can move: if adjusted forecasts for one part family prove consistently better, the plant can give schedulers more room there and less on the families where their changes add error.

I would add one caution to your conclusion. Removing human judgment altogether would throw away the large, information-based adjustments that the study found useful. A supplier that hears a customer is moving a vehicle program to another plant should change its forecast at once. Do you know whether your plant records the reasons for its overrides? If it does, separating information-based changes from routine nudges would show quickly which ones to keep.

Reply to Priya

Priya, your thread on the building products distributor made a point I had not considered: its MRP plans purchases well for steady items such as fasteners, but it struggles with lumber and decking, where customer orders arrive in sudden large batches. You concluded that the distributor needs a better forecasting model. I agree that forecasting matters, but I wonder if the lumpiness you describe is created partly by the ordering process itself, which a better model would not fix.

In a widely read Management Science article, Lee et al. (1997) traced the bullwhip effect, the tendency for order variability to grow as it moves up a supply chain from retail sales to the factory. One cause was order batching. Customers who place orders periodically, or who wait to fill a truck to save on freight, turn smooth consumption into irregular spikes. Another was price variation: when suppliers offer deals, buyers order ahead and then stop ordering, so the orders a distributor sees no longer reflect what end customers are buying. The contractors and lumber yards your distributor serves probably do both. They wait for a full truckload and they buy heavily during spring promotions.

If that is right, the distributor's MRP is forecasting a signal that has already been distorted. A more sophisticated model fitted to lumpy order history would still be fitting the batching rather than the underlying demand for decking on job sites. The remedies Lee and colleagues proposed attack the distortion instead. One is sharing information, such as point-of-sale or job-schedule data from the largest customers, so the distributor plans from consumption. Another is smoothing incentives, for example replacing spring deep discounts with everyday pricing or offering mixed-load freight discounts so customers do not wait for a full truck of one product.

A simple test would show whether batching is the main problem. If the distributor compares its weekly orders from contractors with the weekly sales those contractors report to their own customers, or even with regional housing starts, the variability of the two series should differ sharply when batching is at work. Researchers who followed Lee and colleagues usually measure the effect as exactly that kind of ratio, the variance of orders placed compared with the variance of actual sales. A ratio well above one for decking and close to one for fasteners would match your description and point to the ordering process rather than the forecasting model.

There is also a link to this week's reading on demand management. The chapter distinguishes forecasting demand from managing it, and the distributor has tools for the second: it can schedule deliveries for its large accounts and agree on order windows that suit both sides. Those steps would give the MRP steadier inputs before any change to the forecasting software.

My question is practical. Would the distributor's large contractor customers share their job schedules, or would they see that as giving away competitive information? If they would not, are there a few customers large enough that sharing with them alone would make a difference to the lumber plan?

References

Fildes, R., Goodwin, P., Lawrence, M., & Nikolopoulos, K. (2009). Effective forecasting and judgmental adjustments: An empirical evaluation and strategies for improvement in supply-chain planning. International Journal of Forecasting, 25(1), 3-23. https://doi.org/10.1016/j.ijforecast.2008.11.010

Lee, H. L., Padmanabhan, V., & Whang, S. (1997). Information distortion in a supply chain: The bullwhip effect. Management Science, 43(4), 546-558. https://doi.org/10.1287/mnsc.43.4.546