Luck, Talent and Effort in Business and Career Success

The literature review calls this its most robust pattern: luck’s share is smallest for averaged, repeated, bounded outcomes (annual wages, whether a small business survives) and largest for one-shot, multiplicative, unbounded outcomes (the value of a startup exit, extreme wealth).

What the research shows

• Across 22,004 venture-backed companies, the average founder received $5.8 million at exit, yet “almost three-quarters of entrepreneurs receive nothing at exit and a few receive over a billion dollars” (Hall & Woodward 2010). Judged by the average, founding a startup looks attractive; judged by the typical case, it looks like a losing bet.
• About 55% of venture-backed startups from 1985–2009 were shut down at a loss, and only about 6% returned more than five times the investment; those few supplied roughly half of all returns (Kerr, Nanda & Rhodes-Kropf 2014).
• Random early boosts given in real funding, endorsement and reputation systems led to more success later: arbitrary early luck can compound (van de Rijt and colleagues 2014).

What the simulation adds

The simulation lived the same people’s working lives twice, with the same talent, effort and background but completely new luck. Only 7.75% of those in the top 1% in one life were in the top 1% again, and 18.2% ended below the median. Yet that is still far more than pure chance (1%), because talent raises the odds. “Mostly luck” and “strongly selected on talent” are both true of the very top.

The gap between earnings and wealth comes from the large wealth events in the model (risky opportunities, windfalls and losses), whose settings are assumptions. With all of them switched off, the three traits explain 60.9% of the variation in wealth.

How large is the head start?

• Huggett, Ventura & Yaron (2011) estimate that conditions at age 23 account for 61.5% of the variation in lifetime earnings and 64.0% in lifetime well-being. With other data targets the figures are 53.8% and 57.3%. Initial skills matter most: one standard deviation more raises expected lifetime wealth by about 47%.
• Keane & Wolpin (1997), as summarised by Todd & Zhang (2020), find that about 90% of the variation in expected lifetime well-being is explained by skills already present at age 16. The review treats this spread across models (roughly 50–90%) as itself a measure of uncertainty.

A head start is not the same as merit

On the review’s definition, luck is anything outside a person’s control at the time. Genes, parents’ resources and skills formed in childhood are not chosen. A study of 870 identical and 857 fraternal UK twin pairs estimates that 37–42% of the variation in becoming an entrepreneur is linked to genes (Nicolaou and colleagues 2008). On a strict definition of luck, most of the 60–90% “initial conditions” share is also luck; on a loose one (luck = only events after starting work), luck is about 35–40% of the variation in lifetime earnings. Which definition is right is a philosophical choice, not an empirical one.

Later luck has large, lasting effects

• Losing a job in a mass layoff costs 1.4 years of earnings when unemployment is below 6%, and “a staggering 2.8 years” when it is above 8% (Davis & von Wachter 2011).
• Graduating in a recession causes earnings losses lasting about ten years, larger for less advantaged graduates (Oreopoulos, von Wachter & Heisz 2012).
• A hospital admission at ages 50–59 cuts annual earnings by about 20%, persistently (Dobkin and colleagues 2018).
• Regions that won Spain’s Christmas Lottery saw more new firms, and those firms were larger and survived longer: who becomes an entrepreneur depends partly on random access to money (Bermejo and colleagues 2022).

• Track records help, a little. Founders whose previous venture succeeded had a 30% chance of succeeding again, against 18% for first-timers and 20% for those who failed before (Gompers and colleagues, 2006 working paper; the 2010 published version gives 21% and 22%). The review reads this as a correlation of only about 0.1–0.15 between a founder’s consecutive outcomes. Part of the edge may be perceived skill: suppliers and customers back founders with a track record, so “success breeds success”.
• Experts cannot predict winners. At one large venture-capital firm, the scores of eventual big winners and failures were statistically indistinguishable; even expert screeners explain only about 1% of the variation in outcomes in advance (Kerr, Nanda & Rhodes-Kropf 2014).
• Base rates are harsh. About half of new U.S. businesses are gone within five years (79.6% survive one year, 50.2% five years and 34.7% ten years, for those born in the year to March 2015).

A tension the simulation exposed

When talent’s effect on venture success is set to the literature’s suggested value (a 1.6 times higher success chance for someone one standard deviation above average talent), the model gives repeat success of only 24.5%, not 30%. Matching 30% would need about 2.07 times, but then founders who failed before would succeed only 17.5% of the time. No model in which persistence is purely hidden skill can explain why founders who failed before do as well as first-timers; selection, learning from failure or perceived skill may explain it.

What the research shows

• The average founder age of the 1-in-1,000 fastest-growing new U.S. ventures is 45.0, and “prior experience in the specific industry predicts much greater rates of entrepreneurial success” (Azoulay, Jones, Kim & Miranda 2020).
• People who became incorporated business owners had, as teenagers, higher learning-aptitude scores and self-esteem, and took more risks (Levine & Rubinstein 2017).
• Venture-capital firms show persistence across their successive funds, even using only information available when they raise money (Kaplan & Schoar 2005; Harris and colleagues 2020/2023).

What the simulation adds

With work income, the most talented tenth end with 3.9 times the typical wealth of the least talented tenth and are about 325 times as likely to reach the top 1%. The rank correlation between talent and wealth is 0.49, which means talent accounts for roughly a quarter of the variation in people’s wealth rank. On average the single richest person is at the 90th talent percentile, and is sometimes of below-median talent.

The literature also suggests that the very top performers can be weaker evidence of skill than the next tier, because extreme results often need luck (Denrell & Liu 2012). The simulation never found this where talent could affect wealth: the top 1% were always more talented on average than the next 9%. This shows the reversal is not automatic; it does not show it never happens.

The simulation ran the same people through the same random events twice: once saving 10% of a steady wage, once with no wage entering their wealth. Because the people and the luck are identical, the difference isolates the effect of the income.

Wealth is in years of median entry-level earnings; everyone starts with 1.0. Averages over 1,000 runs of 1,000 people.

A wage is a stream of many small, mostly reliable payments. It adds a large, predictable amount to everyone’s wealth, which dilutes the influence of a few lucky or unlucky big events. And because earnings reflect talent, effort and background, outcomes become more closely tied to those traits.

“Take more shots” cuts both ways

The review advises people to maximise their number of independent “shots” while surviving bad draws. In the simulation, more effort brings more risky opportunities. Without income, going from the lowest- to the highest-effort tenth raises the chance of reaching the top 1% from 0.56% to 1.783%, but lowers the typical outcome from 1.98 to 1.32 and raises the share ending poorer from 23.3% to 41.6%. Extra shots are worthwhile only together with the ability to survive the losses.

The model leaves out debt, living costs and inheritance; its authors say such omissions probably make the version with income too equal and too safe at the bottom.

• The bias is universal but varies. It is smaller in Asian samples (d = 0.30) than in U.S. (1.05) or other Western (0.70) samples, and smallest in samples with depression (0.21) (Mezulis, Abramson, Hyde & Hankin 2004). An effect size d measures a difference in standard deviations.
• Survivorship bias. Studying only successful firms hides the many who did the same things and failed. With about 50% of new businesses gone within five years and about 75% of venture-backed founders getting nothing at exit, success-only samples are badly unrepresentative (Denrell & Kovács 2008).
• Perceived skill can create its own success. When suppliers and customers back founders with a track record, the perception of skill produces persistence, so persistence overstates “pure” skill (Gompers and colleagues 2010).

What this means in practice

The review recommends judging decisions by their process rather than their outcome, using base rates (about 50% five-year business survival, about 75% zero payoff for venture-backed founders), and discounting extreme track records.