La Bourse / Volume VI / Nº VI.08 / Workbook — the Gainshare employee

A team seated around a white table in a meeting, one woman standing to speak, a board of sketches behind them. — Plate VI.08 · Workbook — the Gainshare employeeThe Two Maps on One Table.Every argument about who decides is, underneath, an argument about whether the boundary of the decision matches the boundary of its consequences. Put the two maps on one table and most of the argument goes quiet.

WORKBOOK — THE LUMINOUS GAINSHARE EMPLOYEE

Chapter VI.08 · Federation and Subsidiarity

For the person working inside a gainshare arrangement — where a defined share of verified improvement returns to the people who created it. Assignment is not an abstraction here. It is the thing that decides whether your improvement is countable, whose ledger it lands in, and whether you get paid for it.

WHY THIS CHAPTER IS ABOUT YOUR MONEY

A gainshare has four parts: a baseline, a measure, a share, and a period with a verifier. Every one of those four sits at a level. Somebody decided whether the baseline is set per site or per group, whether the measure is common or local, whether the share pool is yours or pooled, and whether the verifier looks at your unit or at the whole.

Nobody wrote those decisions down as decisions. They were inherited, and they are worth real money to you.

Here is the mechanism, stated plainly. If you improve something whose benefit leaks out of your unit, and the scheme measures only your unit, you have just created value you cannot claim. That leak has a name and a number: s, the spillover fraction. The chapter's whole apparatus is a way of finding it, and in a gainshare it is the difference between an improvement that pays you and an improvement that pays somebody else.

The reverse is also true and less obvious. If your unit's work is measured against a common baseline set for everybody, and your unit is genuinely unlike the others, you are being scored against a standard that does not fit you. That has a name and a number too: h, preference and condition heterogeneity. Where h is high and the measure is uniform, the scheme is quietly taxing the unusual units — which may be yours.

Two numbers. Both estimable in ten minutes. Both worth arguing about with evidence rather than with feeling.

PART ONE — DISCOVERY

Days 1–30: find where your gain actually lands

Exercise 1.1 — Map the four parts to their levels (two hours)

Take your scheme document. For each of the four parts, write the level it sits at and the date it was set.

Part	Level: unit / cluster / group	Set when	Set by
Baseline
Measure
Share pool
Verifier

Most people find at least one blank they cannot fill. A part of your scheme whose level nobody can name is a part that has never been reviewed, and it is the first place to look.

Exercise 1.2 — Find three improvements that already paid (90 minutes)

Appreciative first, and not as a courtesy — as calibration. Name three improvements from the last two years that were made, measured, and paid. For each, write why the measurement caught it. You are learning what a countable improvement looks like in your specific scheme, from the cases where it worked.

Exercise 1.3 — Find one that leaked (90 minutes)

Now name one improvement that was real and did not show up in anybody's statement. There is always one. Write where the benefit went: another unit, the brand, next year, a customer who buys from someone else in the group.

That is your first measured s, and you will use it in Part Two.

PART TWO — THE ARITHMETIC

Days 31–50: compute what is leaking

Exercise 2.1 — Estimate s for your five biggest improvement levers (2 hours)

For each thing you could improve — quality, cycle time, retention, waste, customer outcomes, safety, knowledge — estimate the fraction of the benefit that lands outside your unit. Use a route, and write the route down.

The leakage question. Who enjoys this who did not pay for it?
The migration route. For anything invested in people, s is roughly the fraction who leave over the payback horizon. This is the biggest one for training and it is the same calculation that put schooling above the local level everywhere: about 57 percent of Americans live in the state they were born in, so about 0.43 of the benefit of educating someone lands elsewhere.
The complaint route. When this is done badly, what share of the complaints come from outside your unit?

Exercise 2.2 — Run the rule on your own scheme (90 minutes)

  measuring it at your level costs      ( s / (1 − s) )²
  measuring it at the group level costs h²
  they are equal at                     s* = h / (1 + h)

Worked, and this is the case most gainshare employees find in their own scheme: s = 0.35, h = 0.28. Shortfall 0.35/0.65 = 0.5385. Measuring at your level costs 0.2899 of the value; measuring at the group level costs 0.0784. Threshold 0.28/1.28 = 0.2188. Ratio 3.70.

Read that as an employee, not as a designer. If s for your lever is above s* and the scheme measures only your unit, the scheme is asking you to create value it has agreed in advance not to pay you for. That is not a grievance. It is a design fault with a number attached, and design faults with numbers get fixed.

Exercise 2.3 — Check the other direction (60 minutes)

Now the case where a common measure hurts you. Estimate h across the units your scheme measures — the coefficient of variation in the level each unit would choose, or in what they already achieve. Six branches choosing weekly cleaning frequencies of 2, 3, 3, 4, 5 and 7 give a mean of 4.00, a sample standard deviation of 1.79 and h = 0.45; a single uniform frequency for all six costs 0.2000 of the function's value.

If h is high and your unit sits at the tail, a common target is costing you directly, and you now have the figure to say so.

Exercise 2.4 — The threshold table, memorised (20 minutes)

  h      s*
  0.10   9.1%
  0.20   16.7%
  0.30   23.1%
  0.40   28.6%
  0.50   33.3%

Carry this. It is the whole chapter in five rows, and it is the fastest way to establish in a meeting that you have done the work.

Exercise 2.5 — Do not lean on mobility (30 minutes)

Somebody will say the market sorts this out — good people move to well-designed schemes. Sorting is real and slow. In the United States 7.8 percent of people moved in 2023–24, about 64 percent of movers stayed inside the same county, so 7.8 percent times 36.0 percent = 2.808 percent crossed a jurisdiction: a half-life of 24.3 years. In 1948, at 20.2 percent movers, it was 9.2 years — the mechanism is 2.65 times slower. Your scheme will not be fixed by anyone leaving. It will be fixed by somebody producing a number.

PART THREE — DESIGN

Days 51–75: make the leak countable

Exercise 3.1 — Propose one measurement, not one policy (2 hours)

The move that works is never "change the scheme." It is "add one measurement at the level where the benefit actually lands." One line, one owner, one reporting period. Cheap, reversible, and it produces evidence rather than requiring it.

Write it as four sentences: what is measured, at what level, by whom, and what it will show within two periods.

Exercise 3.2 — Separate the decision from the pool (60 minutes)

Two things get conflated and separating them wins most of these arguments. Where a function is decided and where the gain from it is pooled are different questions. A standard can be set centrally while the gain from beating it is credited locally. A benefit that lands across five units can be credited across five units without anybody losing control of their own operation.

Write your proposal with those two on separate lines. It removes the fear that you are asking for a reorganisation.

Exercise 3.3 — Ask for a cluster, not a merger (90 minutes)

Where a benefit leaks to exactly two or three units and no further, the right boundary is exactly those units — a cluster ledger for one measure, and nothing else. Frey and Eichenberger's idea, applied to a pay scheme: let a boundary form around a single function rather than pushing every function through one boundary.

This is easier to get agreed than anything else in this workbook, because it costs nobody anything and it is obviously reversible.

Exercise 3.4 — Learn the floor, so you can ask for it properly (90 minutes)

Some units cannot perform a function well because they do not have the capacity, not because they do not try. In a gainshare that shows up as the same units missing the same targets, year after year, and eventually as a belief that those units are weak.

The correction is a floor, at a computed height:

      F*  =  100 x ( 1 − C / 2W* )

With C = 120,000 a year to do the function properly and W = 225,000 a year at stake, C / 2W = 0.2667 and F* = 73.33. At that floor, across forty members with indices from 62 to 145, the facility costs 32,000 pounds a year, removes 36,865 of expected loss, nets 4,865, and returns 1.15 pounds per pound of top-up. The twenty members above index 100 hold 399 index points between them; at 80 pounds per index point it funds itself, and the strongest member pays 3,609 a year — 1.60 percent of its own stake, 0.0053 percent of combined turnover.

Two things to take from that as an employee. First, the ask is small, and smallness is what gets it approved. Second, do not ask for full levelling. Levelling every unit to 100 costs 385,200 to remove 176,692 — 0.46 pounds per pound, destroying 208,508 a year. Ask for the floor and you are asking for something that makes money; ask for equality and you are asking for something the finance function will correctly refuse.

PART FOUR — DESTINY AND DELIGHT

Days 76–90: make it hold

Exercise 4.1 — Get the two estimates into the standing template (one hour)

Any proposal to change where something is measured should carry s and h with a route and a name. Once that is in the template it survives everybody currently in the room, which is the only kind of change that lasts.

Exercise 4.2 — Put your name on your estimate (ten minutes)

An estimate with nobody's name on it cannot be corrected, so it hardens into a fact. Put your name on yours. It invites correction, and being corrected on a number you produced is how you become the person whose numbers get used.

Exercise 4.3 — Watch capability move (ongoing)

The variable that moves fastest is what the lowest level can actually do, and it usually moves upward in cost. Rabobank's local member banks fell from 174 in 2010 to 106 in 2015 — 39.1 percent in five years — then merged into one legal entity, because compliance and capital rules had raised the price of being capable. The German cooperative banks went from about 7,100 around 1970 to 672 at the end of 2024, a fall of 90.5 percent over 54 years at a compound 4.27 percent a year.

And the half worth holding onto: through all of that, the functions assigned to their federation did not move. The joint protection scheme has run since 1934, and in the ninety years to 2024 the number of member failures that cost a depositor money is zero. A well-placed function outlives every structure around it.

Exercise 4.4 — Notice what changes in the room (ongoing)

There is a specific relief in the meeting where somebody asks what fraction of the benefit leaves the unit, and a conversation that was about fairness becomes a conversation about an estimate. Estimates can be wrong in public without anybody losing face. Watch for that moment. It is what this chapter is actually for.

KNOW YOUR SCHEME — A CHECKLIST

Answer each yes or no. Every no is a question, not a complaint.

I can name the level at which my baseline is set, and the date it was set.
I can name the level at which my measure is computed.
I can name the level at which the share pool sits.
I know who verifies, and whether they look at my unit or the whole.
I have estimated s for my three biggest improvement levers, with a route.
I have estimated h across the units my scheme compares.
I know whether s exceeds s* for my main lever.
I know of at least one real improvement that leaked, and where it went.
I know what my unit's capability index would be if anybody built one.
I know the review date for the scheme, and it is in my calendar.

Six or fewer yes answers is the normal starting position. Two of them are worth fixing this quarter and the rest can wait.

THE CONVERSATION, SCRIPTED

For the meeting where you raise it. Four moves, in this order, and the order is the point.

One — open on what works. "The scheme caught three real improvements last year and paid them, and the baseline process is clean. I want to talk about one category it does not see."

Two — name the leak with its number. "When we improve handover quality, my estimate is that about 0.35 of the benefit lands in the two units downstream. The route is the complaint log: that share of the complaints about our work come from outside our unit."

Three — do the arithmetic out loud. "Units differ by about 0.28 on this, so the threshold is 0.2188. We're at 0.35. Measuring it only at our level costs 0.2899 of the value against 0.0784 for measuring it across the cluster — a ratio of 3.70."

Four — ask for the smallest thing. "I'm not asking to change the scheme. I'm asking to add one measurement at cluster level for two periods, and then look at what it shows."

That last move is why this works. You are asking for evidence, not for money, and evidence is cheap to grant. The money follows the evidence, which is the correct order and the only durable one.

APPRECIATIVE QUESTIONS FOR YOUR TEAM

Which improvement did this scheme catch and pay properly, and what made it visible?
Where have we helped another unit in a way that everybody knew about and nobody counted?
If the thing we are best at were measured at the level where its benefit actually lands, what would show up?
What would we attempt next year if the units with the least capacity had what our median unit has?
What can our team do now that it could not do two years ago, and which part of the scheme should move because of it?
When we next disagree about whose number this is, what will we point at — and who outside this team could point at it for us?