Surviving Survivorship Bias
Surviving Survivorship Bias
You know about Apple, Microsoft, and Google. You've read their stories—how they were founded in garages, disrupted incumbent industries, and became trillion-dollar companies. You might even believe you can replicate this success by finding the next garage startup.
But survivorship bias blinds you to the 10,000 other garage startups that failed. It blinds you to the survival strategies of companies that seem dominant today but might be tomorrow's Kodak. And it blinds you to the historical returns of strategies that "worked" for the survivorship bias-afflicted winners but destroyed wealth for everyone else who tried them.
Quick definition: Survivorship bias is the tendency to focus on successful entities while ignoring unsuccessful ones that no longer exist, leading to distorted conclusions about causation, strategy, and risk.
Charlie Munger has emphasized this mental model repeatedly because it's one of the most pernicious errors in investing. The very act of looking at current winners introduces a systematic bias that makes you overestimate the replicability of their success and underestimate the role of luck.
Key takeaways
- The dead companies tell the real story: You can't learn what works by studying winners alone. You must study failures and understand why they failed.
- Luck is more prevalent than you think: Many "successful strategies" were actually lucky outcomes. Many failed strategies looked reasonable at the time but hit unexpected headwinds.
- Historical data is biased: If you look only at currently profitable sectors or strategies, you're seeing the survivors of what might have been a much larger pool.
- Index funds carry survivorship bias: Your benchmark might look more profitable than it actually is because failed companies have been delisted and removed from the index.
- Business books amplify survivorship bias: Most business books are written by or about successful people. The failures don't write books—so you learn the wrong lessons.
- Reversion to the mean is real: Companies and strategies that appear dominant today are often revert to the mean precisely because they were lucky outliers.
How survivorship bias distorts decision-making
Example 1: Learning from successful companies.
You read about how Amazon's Day One culture, rapid experimentation, and willingness to take losses drove its dominance. You conclude that willingness to lose money is the path to dominance.
But you never see the hundreds of startups that burned cash with Day One culture and failed. Nor do you see that Amazon's losses were strategic—part of a plan to achieve eventual profitability in a defensible business. Most cash-burning startups are just losing money.
The lesson you should learn is: Amazon was successful despite massive early losses because it had a defensible long-term strategy and enough capital to execute it. Most companies that spend like Amazon just die.
Example 2: Value investing strategies that worked in the past.
You read about Benjamin Graham's success with net-net investing, buying stocks below working capital. You try the strategy today.
But you're experiencing survivorship bias. The stocks that Graham bought 80 years ago that were "below working capital" either (a) worked out and are now famous case studies, or (b) went bankrupt and are forgotten. The thousands of net-nets that Graham looked at but rejected (correctly) never make it into the history books. And the net-nets that he bought and lost money on are downplayed.
Today's net-nets are different. The businesses are in different industries. The capital structures are different. The global economy is different. But because of survivorship bias, you might assume the strategy works the same way it did 80 years ago.
Example 3: CEO characteristics of successful companies.
You read that Steve Jobs was a perfectionist micromanager, and so was Elon Musk, and so was Jeff Bezos. You conclude that perfectionist micromanagement is required for success.
But you never meet the perfectionist micromanagers whose companies failed. You never hear about the brilliant founders whose obsessive attention to detail led to burnout, mutiny, and collapse. Survivorship bias shows you only the micromanagers whose personality trait happened to align with success—not the hundreds whose personality traits led to failure.
Survivorship bias in financial data
This bias extends to financial data itself.
Stock indices are biased toward survivors.
The S&P 500 today looks different from the S&P 500 of 1980. Many of the 1980 constituents went bankrupt, merged, or became irrelevant. They've been replaced by winners.
If you look at the historical performance of the "S&P 500" from 1980 to 2024, you're actually looking at the performance of companies that survived and thrived over those 44 years. You're not seeing the performance of the actual 1980 S&P 500 (many of which failed). This biases historical returns upward.
Delisted companies are hard to find.
If you use a financial database that only shows currently traded companies, you're introducing survivorship bias. The database shows the winners but doesn't track the companies that went bankrupt or were taken private. Getting accurate historical returns requires including the dead companies.
Mutual fund data is highly biased.
When looking at historical mutual fund performance, you see only funds that survived. The funds that lost money and were closed are often excluded from the dataset. This biases historical returns upward. A study by Vanguard found that survivorship bias can overstate mutual fund returns by 1-2% per year.
The dangers of survivorship bias for investors
1. You overestimate the replicability of success.
If 10,000 investors try a strategy and 100 are successful, you read about the 100 and assume the strategy works. You don't see the 9,900 failures. The successful 100 might just be lucky.
2. You underestimate the role of luck.
Many successful business strategies involved huge elements of luck: timing of market conditions, avoiding disruption, being in the right place at the right time. But because you see only the winners, luck becomes invisible. You attribute success to skill.
3. You ignore the costs of failure.
In your dataset (successful companies), the cost of failure is invisible. No one has gone broke. But in the real world, attempting a "successful" strategy and failing costs real money. If the success rate is 1%, but each failure wipes out your capital, the strategy is terrible—even if it worked for the 1%.
4. You become overconfident.
Overconfidence is fueled by availability bias and survivorship bias. You know the success stories. You don't know the failure stories. So you overestimate your ability to identify the next winner.
Recognizing and correcting for survivorship bias
1. Always look at the failures.
When studying a successful company, ask: What would have had to go differently for this company to fail? Look for companies that had similar strategies but failed. This shows you the role of luck versus skill.
Example: When studying Apple's success, study also:
- Companies that tried to make beautiful consumer tech and failed (Jawbone, Path, ColorWare).
- Companies that had great design but poor operations (Motorola Razr had incredible design but failed to compete with iPhone).
- Companies that had great operations but lost to design (Blackberry had superior security and enterprise relationships but lost the consumer battle).
This shows that Apple's success was not inevitable. It required both design excellence AND operational excellence AND timing. Many companies had one or two of these and failed.
2. Adjust historical return assumptions downward.
If you see historical data showing that a strategy or asset class returned 10% per year, assume it actually returned less—maybe 8-9%. Account for survivorship bias by being conservative.
3. Look for data that includes delisted and failed companies.
Use databases that track both living and dead companies. Academic datasets often include delisted companies. Use these for research rather than only using current data.
4. Understand the base rate of failure.
In your industry or strategy of interest, what's the base rate of failure? If you're evaluating startups, know that 90% fail. If you're evaluating small-cap stocks, know what percentage go bankrupt. This grounds your expectation in reality.
5. Seek out the contrarian literature.
Read the books about failures: The Innovator's Dilemma (why successful companies fail), A Few Lessons for Investors by Warren Buffett (discussing his mistakes), Fooled by Randomness by Nassim Taleb (about survivorship bias and luck). These correct the survivorship-biased narrative.
Real-world examples of survivorship bias
The Dot-Com Bubble:
In the late 1990s, people read about Amazon, eBay, and Yahoo's success. They concluded that any internet business was a buy. They didn't see the 5,000 dot-com companies that went bankrupt. Survivorship bias made the internet boom look inevitable rather than a lottery where 99% of tickets lost.
The Japanese Bubble of the 1980s:
Investors read stories about Sony, Canon, and Toyota's success and assumed Japanese companies and the Japanese market were unstoppable. They didn't see that Japanese real estate and equity markets were catastrophically overvalued. The survivors (Toyota) were successful, but the broader market was a trap. Survivorship bias made people overestimate the strength of the Japanese economy.
Peter Lynch's Track Record:
Peter Lynch managed Fidelity's Magellan Fund and achieved 29% annual returns from 1977 to 1990. This became legendary. People studied his methodology, believing they could replicate his success.
But survivorship bias clouds this story. Lynch had the benefit of:
- Starting from a small base (easier to grow percentage-wise).
- Investing during a bull market (1977-1990 had favorable returns).
- Having massive capital advantages (institutional access, research, information).
- Lucky timing and stock picking (he himself admits luck played a role).
Investors who studied his methods and tried to replicate them generally failed because they didn't have his advantages, weren't investing in the same environment, and didn't account for the role of luck. The narrative of "Peter Lynch picked great stocks" is survivorship-biased; the reality is "Peter Lynch had advantages and got lucky."
Warren Buffett's Track Record:
Similarly, Buffett's 20%+ annual returns for 60 years look like pure skill. But consider survivorship bias:
- Buffett started investing in a bull market (1950s-60s).
- He was early to recognize certain advantages (intrinsic value, moats, compounding).
- He had access to capital and investments that others didn't.
- He got lucky on some investments (Berkshire Hathaway could have been an eternal textile mill).
Buffett is clearly skilled, but the magnitude of his outperformance is partly luck and partly circumstances that are unlikely to be replicated. The narrative of "follow Buffett's rules and achieve 20% returns" is survivorship-biased.
FAQ
Q: Does survivorship bias mean that success is just luck and there's no skill?
A: No. Success involves both skill and luck. Survivorship bias just makes it hard to distinguish between them. Over very long periods and large sample sizes, skill usually shows through. But in any individual case, luck plays a bigger role than visible.
Q: How do you guard against survivorship bias in your own investing?
A: (1) Study failures as thoroughly as successes. (2) Assume luck played a larger role than the narrative suggests. (3) Use conservative assumptions about replicating past returns. (4) Recognize that your own successes probably involved luck—don't overestimate your skill.
Q: If 100 people flip a coin 10 times and one gets 10 heads, did they have skill?
A: No—it's pure luck. Survivorship bias would suggest they're skilled if you only interview the person with 10 heads and not the 99 failures. This is a metaphor for many strategies—some succeed by luck rather than skill.
Q: Can you ever escape survivorship bias entirely?
A: No, but you can reduce it by being aware of it. Whenever you see a success story, ask: How many similar attempts failed? What would I be saying if this had failed? This mental exercise helps correct for the bias.
Q: Is the survivorship bias model more relevant for stocks, funds, or businesses?
A: All three. But it's most pernicious for funds and strategies because the selection bias is built into the data. If you're studying a specific company's success, you have more control over your research. If you're studying historical fund performance, survivorship bias is almost invisible.
Related concepts
- Selection bias: A broader category of bias where the sample of data you see is not representative of the full population.
- Availability bias: The tendency to overweight information that is easily available, which is often biased toward survivors.
- Luck versus skill: The distinction between success caused by repeatable skill versus random luck—survivorship bias makes this hard to assess.
- Reversion to the mean: Extreme success is often followed by regression to average performance, partly because extreme success often involved luck.
Summary
Survivorship bias is one of the most insidious mental errors in investing because it masquerades as evidence. You see successful companies, successful strategies, and successful investors and assume you understand what made them successful. But you're looking only at the survivors.
The companies that failed, the strategies that didn't work, and the investors who blew up are invisible to you. This skews your understanding of causation, risk, and replicability.
Munger's insistence on this mental model is a reminder to be humble. When you see a success story, remember the graveyard of similar attempts that failed. When you evaluate a historical return, assume survivorship bias is inflating the numbers. When you study a successful investor, remember that luck was a larger component than the narrative suggests.
This doesn't mean giving up on analyzing success. It means being skeptical, digging deeper, and always asking: What would I be concluding if this had failed? What am I not seeing?