Priya had gotten a bonus in December for six straight years. Eleven percent one year, seven the next, then nine, then nine again. By year five she was quoting her own pay to herself as salary plus about eight.
In November of the sixth year her director told the team there would not be one. Not reduced. None.
Priya is invented. Putting the scene on a made-up person lets me describe what a bonus looks like from the receiving end without describing anybody's real pay.
Most people believe a bonus proves something a raise cannot, because a payment out of the surplus is evidence there was a surplus to share. I wrote that sentence in a book last year.
Here is what the evidence supports, and here the evidence is what a payment can rule out rather than a study. A bonus that has never once been zero cannot separate a company sharing a real surplus from a company protecting a number it has decided not to cut. Six good years look identical under both, and the person receiving the bonus has no way to tell which one she is in.
The bonus is the seen cost and it clears the account every December. The unseen cost is paying for an incentive effect the employee cannot receive, because the signal carries no information she is able to read. The payment goes out at full price whether or not it buys anything.
The sentence I have to withdraw
In a manuscript I am working on, I argue for splitting what is left at the end of a month into pieces with names, and one of those pieces funds payouts to the team. Then I wrote that a bonus paid out of that piece tells the person receiving it something a raise never can, because a bonus is proof there was a real surplus to share and a raise is proof of nothing but a decision to pay more.
I still think that is the best sentence in the chapter. It also rests on a condition I never wrote down, and without that condition the sentence is false.
What a payment can prove
A payment proves what it rules out, and nothing else.
Priya's bonus moved. Eleven percent down to seven is a cut of more than a third, and a fixed cost with a new name on it does not do that. So, the variation on its own already rules out the thing I said the zero was needed to rule out. My sentence was wrong on its own terms before anybody got to the hard part.
The hard part is the second story, and variation cannot touch it. A bonus can be treated inside a company as something owed. In a bad year it then gets found, out of borrowing or out of a cut made somewhere the team cannot see. Six Decembers in a row cannot tell that apart from a surplus that happened to be positive every time.
Priya read her bonus as compensation. That was a reasonable inference from six years of behavior and, as year six proved, the wrong one. She had no way to tell.
The zero year
The natural reading is that the zero year is the year the arrangement broke.
It is the opposite. The zero year is the year the arrangement becomes readable to people who cannot see the books. If employees cannot see the surplus calculation the bonus is drawn from, a run of positive bonuses alone cannot separate the two stories, and the people who can tell are the ones with access to the financial decision.
A company that quietly funds a bonus out of borrowing to avoid a hard meeting has not protected morale. It has spent the one year that would have added information to the six before it. A zero year would not have proved that every prior bonus was a genuine surplus share. It would have shown that the payment is capable of moving to zero, which is the thing the six years cannot show on their own.
There is a catch here I have not resolved. If a bonus becomes informative only after a later zero, then no bonus is informative on the day it arrives, which is the day it is supposed to do its work. I do not have an answer to that.
What I checked
Profit sharing does appear to work, in the modest sense. The best current evidence pools 355 estimates from 56 studies, corrects for the tendency of journals to publish findings that worked, and comes out positive.
Employee ownership is smaller than its advocates suggest. Across 102 samples and about 57,000 firms the correlation with firm performance is 0.04, which is very small. That figure is an association and carries no causal reading.
Neither of those is about what a bonus communicates. They are here because somebody will ask.
The one that does bear on the argument is a gap. For a bonus to prove a surplus, the person receiving it has to be able to see there was one, which is open-book management, and my manuscript never raises it. Having gone looking, I understand why. I found no peer-reviewed evidence base for open-book management at all.
What to do
Ask when you last paid zero.
If the answer is never, that does not prove anything is broken. Priya's employer had never paid zero in six years and may have been running it honestly the whole time.
What it does mean is that nobody in the building can tell, including the people whose behavior you are trying to change. A signal nobody can read is doing none of the work you are paying for, and you are paying full price for it every December.