If you start a business after losing your job, does it go worse?
The field splits founders into two kinds. Pushed, meaning you started because you had no better option. Pulled, meaning you saw something and went after it. The research is mostly built on the assumption that the pushed ones do worse.
They do, on one measure. They do not on another. And the two measures sit in the same table of the same paper.
The table that disagrees with itself
Evidence Norwegian administrative registers followed people whose employer went bankrupt and compared them against founders coming out of stable firms. On personal income the gap is enormous. Founders from bankrupt employers saw a mean change of minus 60,000 Norwegian kroner. Founders from stable firms gained 152,000.
Three rows down in the same analysis, the firms those people built. Mean change in employees was 6.6 for the displaced founders and 5.9 for the others. The share of limited-liability founders who created at least two jobs was 58.9 percent for the displaced against 54.9 percent.
So their companies hired slightly more while their own income fell.
And the gap is one tail, not a worse middle
Evidence That 218,000 kroner difference at the mean is 33,000 at the median. The authors say why in their own words, that the large difference in average outcomes was strongly influenced by a relatively small number of extremely successful entrepreneurs coming out of stable firms.
Report the mean and pushed founders look broken. Report the median and you have two similar distributions with different upper tails. Both are correct. Only one of them gets quoted.
The best experiment says the quality does not drop
Evidence France changed its unemployment rules in 2001 so that people could keep topping up benefits against business income instead of forfeiting them. Researchers used it as a natural experiment across more than a million firm-level observations.
Firm creation rose about ten percent. Every test of whether the new entrants were worse came back null, and the precision is the point. The base rate for closing within two years is 16 percent, and the data rules out the new firms being worse by more than 0.6 percentage points. On hiring, with a base rate of 25 percent, the worst case the data allows is 2.5 points.
Their sentence is that the reform significantly increased firm creation without decreasing the quality of new entrants.
Where the pushed founders really do worse
Evidence It is not nothing. Using an objective definition, unemployed in the period before starting, American data shows new necessity founders are about half as likely to incorporate, 10.5 percent against 19.1, and far less likely to have employees, 4.6 percent against 14.9. German data shows 75 percent working alone against 53 percent, and monthly labour income of 1,521 euros against 2,536.
Those are unconditional averages, not causal estimates, and the authors say so.
The number that breaks survival as a measure
Evidence Swedish registers followed 271,873 people displaced by plant closures against 200,000 matched controls, for three years before and eleven after. The displaced who went into self-employment were less likely to leave it. Eleven years on, eight percent were still trading against three to four percent of the non-displaced.
The channel that matters is exit into unemployment, which is what failing looks like in this data, and displacement had no effect on it at all.
The researcher does not read longer survival as better businesses. She reads it as having worse alternatives, self-employment becoming a means of subsistence rather than a stage. The people stayed because leaving was worse.
That is the finding worth carrying out of this. The same number, longer survival, supports opposite conclusions depending on what you know about the person’s other options. A dataset that records only the business cannot tell you which story is true.
The assumption underneath the whole question
Evidence Researchers asked people what they wanted at the moment they were founding, rather than afterwards. About three quarters said they wanted to stay small. The median expectation was three or four employees at year five. Nearly 80 percent said research and development would not be a priority. More than half gave a non-pecuniary reason as primary, being their own boss, controlling their schedule, working from home, against about a third citing income.
Very few said they had no better job options.
So a small business is not evidence that somebody was pushed. Small by choice is the ordinary case across every origin, which breaks the inference most writing on this topic makes without noticing it is making one.
The honest summary
Person-level income differs a lot and is driven by a thin right tail. Firm-level outcomes barely differ. Survival differs in the wrong direction and means something other than it appears to. Whether the pushed founders do worse depends on which of those you were asking about, and almost nobody says which.
Where this stops. France, Norway, Sweden, Germany and the United States, all rich-country administrative or panel data. Necessity is defined throughout as having been unemployed beforehand, which is narrower than needing money. Several key papers are paywalled and I read working-paper versions.
What I am not saying. That losing your job is a good way to start a business. The income evidence is real. The claim is that the firm outcomes and the person outcomes point different ways, and that survival is not the quality measure it is usually taken for.