Named by James Alfred Ewing for magnetic materials (1881) and carried into economics by Edmund Phelps (Inflation Policy and Unemployment Theory, 1972) and by Olivier Blanchard and Lawrence Summers (Hysteresis and the European Unemployment Problem, 1986).
Recognizes when a system's state depends on the history that produced it, so reversing the cause will not reverse the effect.
A price rise loses customers to a rival, and taking it back does not bring them home
Our price sits 20 percent above the rival's from month 2 to month 7, then comes back down. Play: customers leave while it is high, and once it falls only those free to switch come back; the ones who signed two-year contracts with the rival stay gone. Set the lock-in to zero and Play again: without it, the loss reverses. Illustrative.
Our price premium
Our customers
Month 24 of 24
- Our price premium
- 0%
- Our customers
- 767
- Switchers locked in with the rival
- 233
- Our price premium
- How far our price sits above the rival's. Customers shrug off up to 5 percent; above that a share of them shop around each month.
- Customers leaving
- Accounts that move to the rival while our price is high: 4 percent a month for every 10 points of premium above 5.
- Switchers with the rival
- Accounts that moved. Those on a two-year contract stay there whatever our price does; the rest drift back once our price is competitive.
- Our customers
- Business accounts on our service: 1,000 at the start.
- Customers returning
- Switchers who are free to leave the rival come back at about a quarter of them a month once our premium is gone.
- Rival's two-year contracts
- The share of switchers who signed a two-year contract with the rival. This is what makes the loss stick: the path the customers took changed the market.
An example
A regional waste hauler serving restaurants and offices raised its rates 20 percent above its main rival's when fuel costs jumped, planning to roll the increase back once fuel came down. Over six months it lost about 310 of its 1,000 accounts. When fuel fell, it cut rates back to match the rival and budgeted for the accounts to return within a quarter. A quarter later about 45 had come back.
The sales director called the accounts that had left. Most had signed two-year contracts with the rival, often with a discount for committing, and a few had no complaints about the switch. The fuel spike was over, and the market it had left behind was a different one. Finance rebuilt the forecast around the contract dates, planning on about 770 accounts until the first contracts expired, and sales built a win-back offer timed to each customer's renewal month, with a credit toward the rival's termination fee.
Why it works
Most plans assume a system will retrace its steps: raise the price and lose customers, lower it and win them back. That holds only when nothing about the system changed on the way, and often something did. Customers who switched signed contracts, learned a new supplier's systems, or built a relationship, and each of those raises the cost of coming back. The state of the market now depends on the path it took, and the same price that held 1,000 accounts before the rise holds about 770 after it.
Asking whether reversing the cause would reverse the effect catches the trap before the plan relies on it. When the answer is no, the useful question becomes what the shock changed, and the remedy works on that change, such as contract dates and switching costs, more than on the original cause.
The thinking behind it
James Alfred Ewing James Alfred Ewing, a Scottish physicist and engineer, studied how the magnetization of iron lags behind the force applied to it while he was teaching in Tokyo, and coined the word hysteresis for that lag. His 1881 paper in the Proceedings of the Royal Society of London reported the work. Raise the magnetizing force and lower it again, and the iron does not return along the same path: its present state carries its history. The term later passed into other fields, economics among them.
Edmund Phelps Edmund Phelps, who received the 2006 Nobel Memorial Prize in Economic Sciences, introduced hysteresis to unemployment theory in Inflation Policy and Unemployment Theory (1972). A long spell of unemployment, he argued, can become partly permanent as workers lose skills and grow discouraged, so a recovery in demand does not restore the old level of employment by itself. The evidence for the effect is mixed across countries and periods, which is a reason to test for it case by case.
Olivier Blanchard Olivier Blanchard, a French macroeconomist at MIT and later chief economist of the International Monetary Fund, set out the hysteresis hypothesis with Lawrence H. Summers in Hysteresis and the European Unemployment Problem, published in the NBER's Macroeconomics Annual for 1986. Temporary shocks, they argued, can raise unemployment permanently, because those still employed bargain wages up instead of making room for those out of work, and because skills fade during long spells without a job.
How to run it
- Ask whether undoing it would restore it. Put the question to the plan directly: if we reverse what we did, do we get back to where we were?
- Find what the shock changed. Contracts signed, skills lost, suppliers switched, habits formed, or people who left.
- Ask the people who moved. Call customers or staff who left and ask what it would take for them to return.
- Size what will not come back by itself. Estimate how much of the loss is locked in, and until when.
- Plan on the new state. Rebuild forecasts and budgets from today's level instead of the old one.
- Act on the path. Work on what holds the change in place, such as contract dates or switching costs, timed to when each loosens.
- Price the exit before the next shock. Count the cost of undoing a temporary move along with what it earns while it lasts.
Where it goes wrong
The usual failure is planning the reversal on the old map. The price comes down, the forecast assumes the customers return, and the budget misses for a year. Spot it when a recovery plan offers no reason the people who left would come back, and recover by finding out what each of them signed or changed.
The second failure is seeing hysteresis everywhere. Plenty of systems do retrace their steps: a seasonal dip recovers, a shift pattern returns to normal. Test for a lock-in before assuming one. The third is missing it among staff. People laid off in a downturn take jobs elsewhere and build other skills, and hiring them back later costs far more than the layoff saved.
When to use something else
- Path Dependence Analysis. When the question is how a long history of small choices locked in today's options, beyond one shock and its aftermath.
- One-Way and Two-Way Doors. When the decision is still ahead and the task is telling which moves can be undone cheaply.
- Second-Order Thinking. When the team needs to trace what a temporary move will set off before making it.
Where it sits in SOLVE eX
Go to the source
- Hysteresis in magnetic materials, the term's first use, Proceedings of the Royal Society, 1881
- Inflation Policy and Unemployment Theory, 1972
- Hysteresis and the European Unemployment Problem, NBER Macroeconomics Annual, 1986