Bullwhip-effekten

The bullwhip effect – why small ripples in a shop become crises in the factory

Why a small change on a shop shelf becomes a large swing in the factory: the bullwhip effect explained, with the nappies that exposed it, the four causes, the toilet paper of 2020 – and what actually dampens it.

Back to the episode Arrow keys or swipe

What this covers

  1. What the effect is, and where the name comes from
  2. The nappies that exposed it
  3. The four causes
  4. Safety margin on top of safety margin
  5. Ordering by the pallet
  6. The promotion that looks like demand
  7. Ordering more than you need, because everyone else does

What this covers (cont.)

  1. What actually dampens it

The nappies that exposed it

Procter & Gamble found that nappy orders from distributors swung more than consumption could explain.

Why it is such a good example: nappy consumption is about as predictable as anything gets. The number of babies does not change from week to week, and a baby uses roughly the same number every day.
Yet the orders swung sharply. The variation did not come from the market. It was created inside the supply chain itself.
From the same paper: at Hewlett-Packard, resellers' orders to the printer division swung far more than customer demand – and the orders passed on to the integrated circuit division swung more still.

The four causes

Four causes, and how they look in a retail chain

CauseHow it looks in a grocery chain
Forecasting on forecastsEach tier builds its forecast from the orders of the tier below – not from actual sales.
Ordering in batchesYou order whole pallets or whole lorries, not what you sold.
PromotionsThe price swings, and purchases move in time.
Gaming a shortageWhen something runs out, everyone orders more than they need in order to get enough.
That all four are rational is why the effect is so hard to shake. You cannot remove it by asking people to stop being stupid. Every tier behaves sensibly given what it can see – the problem is that it cannot see enough.

Margin on margin

The first cause is the most underrated, and it can be calculated.

Ten per cent extra, four times over

  1. The shop sells 100Actual consumption. The only real number in the whole chain.
  2. The shop orders 110A little extra, in case it picks up.
  3. The wholesaler orders 121Sees 110, adds ten per cent.
  4. The manufacturer plans 133Sees 121, adds ten per cent.
  5. The raw material supplier gets 146Forty-six per cent above what was actually sold.
The arithmetic is 1.1 to the power of four, or 1.46. Nobody has done anything wrong. Every tier added a perfectly defensible margin. But the margins multiply, they do not add – which is why the amplification gets so strong with only a few tiers. And if sales drop again next week, the whole chain runs in reverse with the same force.

Ordering by the pallet

The second cause is physical, and it is built into the way goods move.

What the supplier sees

What happens in the shop

Steady sales

  • Four packs Monday
  • Four packs Tuesday
  • Four packs Wednesday
  • Entirely predictable

What the supplier sees

Nothing, nothing, a whole pallet

  • Nothing for several days
  • Then one large order
  • Then nothing again
  • Looks like violent variation
From the supplier's side this is indistinguishable from genuine swings in demand. And if many shops share an order day, all the batches land at once. The supplier then gets a peak that does not exist in reality – it was made by the calendar, not by customers.

The promotion that looks like demand

The third cause is the most recognisable in grocery retail, and it is self-inflicted.

The promotion cycle

  1. 1PromotionThe price drops. Sales shoot up.
  2. 2Customers stock upPeople buy three packs instead of one, because it pays to.
  3. 3The shop fills upAnd the chain orders heavily into the promotion.
  4. 4Afterwards: silenceCupboards at home are full. Sales fall below normal.
  5. 5The forecast is confusedThe next promotion is planned on figures that contain the last one.
The decisive point is that a promotion rarely creates new consumption. Nobody eats more crispbread because it was on offer – they buy what they would have bought anyway, earlier and cheaper. The peak is therefore purchases moved in time, not growth. But in the sales figures the two look identical, and that is where the forecast breaks.

Gaming a shortage

The fourth cause is the one that makes crises worse, and it is the most human.

The toilet paper of 2020 is the textbook case, and it was not about anyone running out of paper.

What actually changed: people were at home around the clock instead of at work and school. Consumption shifted from bulk packs in workplaces to ordinary packs in shops. Total consumption changed little.
What set off the crack: empty shelves were filmed and shared. People bought two packs instead of one. Shops ordered extra. Chains ordered extra on top.
Why it is bullwhip and not merely panic: every tier was reacting to the orders from the tier below, not to consumption. And those orders already contained everyone else's fear.
And the other side: when it settled, the chains were left with stock they could not shift, because customers had full cupboards.

What actually dampens it

Here is the most important insight, and it runs against instinct.

Two ways to meet the problem

Works badly

Guess better at each tier

  • A better forecasting model for each party
  • Larger safety stock to absorb the swing
  • Sharper buyers
  • The problem: everyone is still guessing from the wrong number

Works

Share the real number

  • Point-of-sale data shared upstream
  • Everyone plans against actual sales, not against orders
  • Smaller, more frequent deliveries instead of large batches
  • Steady prices rather than large promotions
  • Allocation based on historical sales, so over-ordering does not pay
The point is that bullwhip is not a forecasting problem. It is an information problem. As long as each tier only sees the order from the tier below, it is planning against a number that is already distorted – and then it does not matter how good the model is. Walmart is the best-known example of the opposite: sharing checkout data upstream with suppliers, so that everyone sees the same thing.

The link to lean

Finally, a nuance worth carrying, because the two subjects are closely tied.

What is worth taking with you

That the bullwhip effect is orders swinging more than sales, and the swing growing with every tier upstream.

Key points

  • The bullwhip effect is that orders swing more than sales, and the swing grows upstream.
  • Procter & Gamble found that nappy orders swung more than babies could explain.
  • Lee, Padmanabhan and Whang identified four causes – and all four are rational choices.
  • Every tier adds its own safety margin, and the margins multiply.

Key points (cont.)

  • Promotions create a demand peak that is purchases moved in time, not new consumption.
  • In a shortage everyone orders more than they need, making the shortage worse.
  • What dampens the effect is shared information, not better guessing at each tier.

Questions this answers

  • What is the bullwhip effect?
  • Why do the swings grow further up the chain?
  • What was the Procter & Gamble example?
  • Why do promotions amplify the effect?
  • What happened with toilet paper in 2020?
  • How can the effect be dampened?

Questions this answers (cont.)

  • What does this have to do with lean?

Sources

  • Hau L. Lee, V. Padmanabhan and Seungjin Whang – Information Distortion in a Supply Chain: The Bullwhip Effect, Management Science 43(4), 1997
  • Jay W. Forrester – Industrial Dynamics (1961), where the phenomenon was first described

Spørretimen

Read the whole thing at

www.sporretimen.no/en/episodes/bullwhip-effekten/