Forrester on amplification in supply chains
📚SOURCE
💎PERMANENT NOTES
🧵SYNTHESIS
In 1958 Forrester did something that now looks ordinary and then was not: he wrote down a company as a set of stocks, flows and delays and let it run. His model has a retailer, a distributor and a factory. Each holds inventory, each takes a week or several to process an order, ship it and adjust. Then he raises retail sales by a tenth and holds them there.
The result is the paper. Orders at the distributor rise by more than a tenth, orders at the factory by more again, and factory output overshoots by several times the original change before swinging below it. The system takes more than a year to settle. Nobody in the model is foolish. Each level refills its own shelves and the pipeline feeding them, which is prudent, and each reads the orders arriving from below as demand, when they are partly demand and partly somebody else's refill.
His claim is that the structure produces the behaviour. Managers blamed the swings on the market or on seasons, and he shows the same swings appearing with the market held flat.
✨KEY IDEAS
A single ten per cent step in retail sales becomes a much larger swing in factory production, months later.
Every level orders to replace what it sold, to rebuild stock, and to fill the longer pipeline, all at once.
The oscillation continues long after the original change has stopped.
Speeding up clerical handling of orders helps less than expected; changing what information each level sees helps more.
He wants managers to design the information and decision rules of a firm the way an engineer designs a control system.
📝READING NOTES
Paraphrase, not quotation. Yusuf Demir sent this in September with a numbered list of reasons to read it. I got to it on 25 and 26 October.
What he built
- Three levels, each with a stock and a delay: time to notice, time to decide, time to ship, time to make.
- Ordering rules taken from how companies said they ordered.
- One disturbance, then nothing. He watches what the structure does with it.
Why it matters to me
It is the origin of the 🧩 Bullwhip effect, forty years before the name. And it is the cleanest case I know of 💎 Delays make a sensible reaction look like the wrong one: the factory manager who adds a shift is right on the day and wrong by the time the shift is trained. I had this from 📖 Meadows on stocks, flows and delays as a principle. Forrester is the worked example, and Meadows worked in his group at MIT, which I had not put together.
For freight it is literal. A forwarder's bookings are orders two or three levels from the shopper. 💎 Demand swings grow as they travel away from the customer is the note this became.
My quarrel with it
- It is a model, and he is very sure of it. The paper promises that simulation will change management as a profession. Parts of that happened. Most firms still order the way his did.
- Nobody in the model talks. No phone call from the retailer saying the rise is permanent. Real chains have people in them, and some of the damping he wants from better rules comes from a planner who knows another planner.
- He treats the swing as pure cost. Some of that stock is the system's slack, and I am not sure all of it is waste.
Next: the 1997 paper that named the effect, also from Yusuf.
Reading record
- Sent by: Yusuf Demir, by e-mail, September
🗣️IN SHORT
- In : Industrial Dynamics: A Major Breakthrough for Decision Makers, Harvard Business Review, 1958
- Forrester argues that : swings in production come from the delays and ordering rules inside a supply chain and not from the market
- by stating that : a simulated chain of retailer, distributor and factory turns one small, permanent rise in sales into a large and long oscillation upstream, with every level acting sensibly