MIT OpenCourseWareSession 7: Market Growth with Boom and Bust
MIT OpenCourseWare
Model scope 0:00
This week’s model combines market growth with a durable product, so demand comes from first-time buyers and replacement purchases. Sales run through a direct sales force, and its productivity depends on delivery delay and on how many customers are still available. The production side includes a manufacturing operation and a backlog of ordered products, but no inventory. The discussion also points to excess delivery delay as an important feedback.
Ways to split it 3:01
The main advice is to break the model into clean chunks. One chunk can handle production with exogenous orders flowing into backlog and out as deliveries. A second can handle the sales force, where available customers and perceived delay shape its effectiveness. A third can handle the market side, where customers become available or move through purchase states. Several student models are then reviewed as examples of these choices.
Customer flow and backlog 22:00
The model separates current customers, potential customers, and the pitching step, so acceptance and sales effort do not get mixed together. That helps isolate delays that come from delivery rather than from the sales force. The discussion also turns to a chain of backlog, work in progress, finished goods, and delivered goods. That adds detail on how units move through the system and where shipment delay enters.
Capacity and delay logic 25:30
The main critique is that capacity is being used as units on the floor, not units per year, so it is really work-in-process capacity. A cleaner version would keep production capacity as a rate and use a separate processing time. The same issue appears in delivery delay and reputation. The desired delivery delay is hidden inside the formula, and the factory construction rule double counts sensitivity and build time.
Sales force limits 43:02
Sales outreach is treated as a gate on conversion. The instructor notes that the model uses a maximum yearly outreach per salesperson and a minimum function to cap actual outreach. He accepts the structure, but says the utilization rule is too sharp. If interested demand is only a little below capacity, the model cuts effectiveness too much. A softer utilization function would be better.
Manufacturing capacity details 47:30
The same concern comes up in the factory model. A factory takes years to build, so a better setup would track work in process and use higher order delays for both construction and depreciation. The current first order treatment is a simplification. He also says the stock-management logic needs replacement inflow as well as the gap to desired capacity, or the model will leave a steady state error.
Capacity and delay logic 1:02:02
Hazhir Rahmandad walks through ways to model expansion from pressure, backlog, and delivery delay. He says a pressure-based anchor can be legitimate, but it still needs a replacement piece or the system reaches a steady-state problem because outflow keeps draining capacity. He prefers the cleaner setup where desired production capacity comes from the expected customer order rate, then adjusts for how long it takes to add capacity. That makes the target closer to what customers will actually need a few months or years ahead. He also notes that anchoring to backlog or recent orders is less prone to wild swings, while stronger reactions can overshoot and create unstable market growth behavior. A backlog divided by delivery delay can be a useful average signal, but when demand is noisy the distinction matters more.
Delay initialization fix 1:23:03
The discussion clears up a subtle mismatch in the capacity delay. If you use DELAY3 without an initial condition, Vensim starts it in equilibrium, which can leave production capacity and its hidden stocks inconsistent. The fix is to set the delay’s initial condition from initial PC divided by facility life so the stock and the delay begin from the same level.
Capacity limits and sales 1:26:00
The model then moves through desired PC building, adjustment for WIP, and capacity backlog. The main concern is that excess delay is floored at one, so desired capacity can never fall below current capacity. That blocks reductions when demand drops. The same section notes the sales side: customer conversion depends on pitches, productivity, and excess delay, but the formulation can make conversion unrealistically high if delay gets too small.
Anchoring capacity changes 1:43:01
The discussion focuses on how to model factory capacity and sales force without losing track of scale. The main warning is against building an authorized factory stock from a pressure signal alone. If the desired number is not tied to current factories or current demand, the model becomes a kind of dead reckoning. It may move up or down, but it does not know whether it is still near the right level. The same problem appears in the delivery-delay feedback. A rule that simply expands when delay is above target and stops when it is below target is too blunt. A better choice is a ratio of actual delay to target delay, anchored to the current operation. That way, if the business grows from one factory to many, the adjustment can also grow instead of staying stuck at half a unit or one unit per month.
Sales force sizing 2:06:01
The discussion focuses on how to set sales force size from the model. A simple budget share is useful at first, but it should change as the market changes. Early on, you may spend more on sales. Later, when customers already know you or the market is mostly covered, that share should fall. A better rule is to link sales spending to the net or marginal value created by salespeople.
Marginal value and diffusion 2:12:03
The key distinction is between average and marginal productivity. Average sales per person can hide a bloated staff. Marginal value asks what one more salesperson adds, and that should drive hiring or cuts. The session closes by moving to next week’s endogenous diffusion model, using pacemakers as an example of a product that improves over time, expands its market, and can also bring bugs.
Adoption delays and bugs 2:20:00
Discovery and fixing of new problems can shape adoption. In pacemakers, delays in finding side effects can create an oscillation: excitement rises, early research finds disappointing results, adoption falls, fixes follow, and then adoption recovers. Word of mouth can spread both the benefits and the bugs.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.
