The wage nobody named as a price

Wage

Two numbers get filed under what the founding cost, and they get filed as one. The first is money that went out. Machines bought, software paid for, months of runway spent before anything came back in.

The number that went out leaves receipts, and a person can produce it on request. The second is money that never came in: the pay somebody else would have put in my hand for the same hours, in a place with a payroll. Nobody sends an invoice for that.

It comes out of the same life as the first, and it goes out on the same days.

One of the two has a name people say out loud. It is the smaller one.

A wage that is never taken is still paid.

Most of what a founding costs is ordinal. Attention, capacity, energy: I can say one year cost more than another and I cannot put a figure on it, and I keep the word cost for that class. The word price I keep for the places where a cardinal figure exists.

The wage forgone is the one cost in the whole founding that has a figure of its class sitting ready, a number some payroll would have printed every month, and it is the one that nobody writes down. The money out gets a spreadsheet. The wage gets a mood.

So the term I would put at the front of any founding, before the first client and before the first machine, is one line: the wage forgone gets stated as a plain figure of its class, on the day the thing starts, and it sits on the page next to the money going out. The figure, written that way, stops being a mood about a hard few years. It becomes the price of something, and a price can be looked at and refused next year.

Left off the page, the same amount leaves anyway. It leaves on the same days and out of the same life, and nothing anywhere carries the figure.

Autonomy

The person who pays that wage buys something with it, and autonomy is the word for what they buy. A person can hold a word like that for years and never once look inside it. Nothing forces the look.

What sits in there stays undescribed, so it never gets set against the wage going out for it, and the paying runs on undisturbed for another year, and then another, and the thing being bought is never asked to show what it is.

I sat down with the arithmetic one evening. On one side, what got built. All of it real, all of it working.

• Brands for other people’s units, running, with their names on every surface.
• Systems inside other people’s companies, still in use, still behaving the way I set them up.
• Drafts, decks, pricing, process, delivered and living inside somebody else’s thing.

On the other side, nothing that carries my name on its own. Not a system I hold, and not one artifact where the work and the person who made it sit on the same page.

That is what the arithmetic returns.

The brochure version of autonomy is freedom to say yes. Anything can be started, nobody’s approval is needed, the day takes whatever shape the work needs. For a founder at capacity that stops being true, and the reason is mechanical.

A thing does not sit in the head as one line on a list. It sits there live, with everything hanging off it, and it keeps running while it sits. A head holds a few of those and then it holds no more.

Somebody already carrying that much gets no gift from a new option. It lands as one more node on a graph with no room left on it, and the node has to be read and fitted against what is already standing before anything can be done with it, including throwing it away. That is the part nobody counts.

Refusing costs nearly what accepting costs, out of the same capacity, so the freedom that got bought turns out to be an obligation to process whatever shows up.

An option handed to a full head is unpaid work.

Nobody sold it in those terms.

In a job the refusals belong to somebody else. A seat above yours holds the whole supply and filters it, and what comes down is a short list instead of the whole queue. That filtering is a real service and it is already inside the wage.

Nobody itemises it, so from inside the job the wage looks like the figure at the bottom of the slip and nothing else. The service only becomes visible after it stops arriving, which is late.

So the question I have been carrying and still cannot answer cleanly. Was I buying the autonomy, or was I buying the right to stop asking a person above me for permission, which is a smaller object and might have come cheaper if anybody had ever priced it on its own?

Price

For a long time I read my thin years as a report on me. Bad timing, bad handling of my own week. Then I went and read what the field had measured, because whether founding underpays is not an open question; it was settled with data long before I entered.

The Decade Hamilton Measured

Hamilton, in 2000, whose study the Journal of Political Economy ran, took people working for themselves and set them beside wage employees of the same ability, then followed both groups across ten years. The person standing in the middle of the self-employed group came out roughly a third under what that same person would have drawn on a payroll. The middle of the whole distribution sat below a salary for a decade, not one bad quarter and not one venture that taught a lesson.

After that paper the field stopped arguing about whether the gap exists and moved to why anybody goes in at all.

The Shape Astebro Found

Astebro, in 2003, then gave the gap its shape. Returns to founding are not a bell curve sitting a little low; they are a lottery. Most draws land under a salary, a very few land enormous, and those few drag the average upward while the person at the median sits quietly below a job.

Read the mean and founding is a reasonable bet. Read the median and it is a poor one for almost everybody who takes it. They are both right, and they are counting different people.

And still the entrants keep coming, in the same volume, year after year, and no correction arrives from the data.

What the Owners Said They Wanted

Hurst and Pugsley, in 2011, asked the question that turns the gap from a loss into a purchase. They went to the owners of small firms with survey data and asked what those owners actually wanted from the business. They mostly had no plan to grow and had not had one at the start.

Few had brought anything new to a market. Few meant to hire past a handful of people. They gave reasons for starting that were not money at all: no boss above them, the hours, the specific work itself.

The paper’s name for this is the non-pecuniary benefits model. What it says is that the owner is spending the earnings gap on something that is not earnings, and spending it on purpose. It follows that most of what gets written about failure measures these people against a target they did not hold.

A firm that holds at one person for years and feeds its owner shows up in the data as a firm that did not grow, and to the owner it shows up as the arrangement working.

Two pieces of hygiene the field supplies on its own. Levine and Rubinstein, in 2017, split the self-employed by whether they had incorporated, and the two halves behave like separate populations: the incorporated out-earn comparable employees, the unincorporated fall behind, and the pooled figure that says founding underpays is partly an artifact of mixing them. And the owners in the survey were asked after the fact, so some of the never-wanted-to-grow answers are a story told to fit an outcome that had already arrived.

The reference class is narrower than the headline. The headline still holds for the middle of it.

The Option and the Report

Manso, in 2016, reframed the whole decision as an option, not as a job. A founder pays to find out something about a bet nobody can judge from outside, and keeps the right to walk away. Measured that way the numbers move.

People who tried it, got their answer, and went back to a payroll inside two years show no lasting penalty on their earnings. The gap in the cross-section is made almost entirely of the people who stayed in after the experiment had already reported. That puts the failure somewhere other than where the folk version puts it: in staying past the point where the answer arrived, not in entering or leaving it.

What I Took from It

I took one word out of it and the word is price. Loss is the wrong name for the same thing, because loss is what happens to a person and a price is what a person pays. Something was bought, this is what it cost, and once the sentence is built that way it can be checked against the year: did the autonomy arrive, and is it worth what it is charging me now.

Then the harder part, which took longer to sit still with. The years of effort with nothing under my own name were not a business failing to appear. They were the purchase going through, every day.

The desk, the shape of the week, the title on the card: that is what the wage was buying, and the daily work was the payment still running. The seat was worn, not held. Nothing in the arrangement was broken.

It was optimising for autonomy instead of for a viable unit, and it was getting exactly what it optimised for. The deficit was not evidence about me. It was the sticker.

Finding that true cost me a story I liked. Unlucky is comfortable and a price is not, because a price means somebody agreed to it. I did, every year, without ever saying the size of it out loud to myself or to anyone else.

You could have been earning proper money all this time.

Foregone

The wage runs whether or not anybody names it, and nobody stops it when the founding stops, because the years it came out of were years when the pay sits high and a job is also compounding things behind your back that nobody counts until much later.

Manso’s failure mode and the unnamed wage are one fact seen from two sides. An experiment can only report if somebody set the terms for what a report would look like, and the price is the first of those terms. A founder who wrote the wage down on day one has a figure the year can be set against, and a figure gives three moves a mood never gives: refuse the next year at the same rate, renegotiate what the seat is charging, or walk.

A founder who carries the same wage as bad luck has nothing for the year to report against. Nothing was agreed, so there is nothing to withdraw from, so the year gets called a rough stretch and the next one goes in on the same terms, and the one after that. The founder who stays past the report has usually done nothing more than this.

Not stubbornness. A missing figure.

Get this wrong and what goes is not the pay. The pay goes under either reading. What goes is the ability to decide anything about it, and I did this for years, and I could not have told you the size of the wage I was spending, because I had never once put it anywhere I would have to look at it.