(Image sources: 1, 2)

Recently, there has been a lot of talk about a possible bubble in startup valuations. The question I want to explore is: “How are VC investment returns affected by bubbles?” That is, if it’s a bubble, should VCs (and by extension, startups that need VC funding) close up shop for a few years and wait until market conditions change?

Assertion #1: VC returns are governed by power laws

Most VC funds have somewhere between 15 and 50 portfolio companies. It turns out that almost all of a fund’s returns will come from its best 1-3 investments. Here’s a chart of outcomes for a typical individual investment:

(Source: Returns to Angel Investors in Groups)

Assigning some sample numbers, this chart suggestions that:

• 52% of investments might each return 0.2X (i.e. only 20% of the initial investment)

• 35% of investments might each return 1.5X

• 8% of investments might each return 6.5x

• 2% of investments might each return 15x

• 3% of investments might each return 40x

This distribution results in an expected multiple of 2.65X for the overall fund (pretty close to the 2.6X mentioned in the upper-right of the chart).

The noteworthy bit is that for a typical portfolio, 5% of the investments will be responsible for more than half of the total returns, and 13% of the investments will be responsible for more than 3/4 of the returns. This is what people mean when they say that VC returns follow the power law: the bulk of the returns are to be found in the outliers, not in the most common outcomes. If a fund invests in Twitter or Nest early on, the fund will be hugely successful; if none of the investments becomes a $1b+ company, then the fund will be mediocre at best. Trying to eke out an extra 1.5X or 6.5X exit is relatively meaningless when a 40x exit could return the entire fund. That’s why investors and founders swing for the fences.

Assertion #2: Investing in a power law distribution is only worthwhile if you might find an outlier.

A typical VC fund makes investments for 3-4 years, then holds onto those investments until every company in the portfolio either goes public, gets acquired, or fails. If outlier companies (i.e. the Twitters and the Nests) predictably occur in some years and not others, then it would be prudent to only invest in the good years. For example, if it turned out that 95% of all $5b+ companies were started during a recession, then making venture capital investments during economic booms would be unwise.

Assertion #3: Great companies are founded every single year

I spent some time trying to dig up the most successful companies founded in each year since 1997 (the year that the Dot-com bubble began according to Wikipedia). The most notable companies that I found are listed below.

(Note: the list is not meant to be exhaustive.)

Year  Company          Valuation**
1997  Priceline           $59b
      Netflix             $40b
1998  Google             $370b
      VMWare             $37b
1999  Salesforce          $47b
      Jawbone            $3.3b
2000  TripAdvisor         $13b
      Pandora            $3.4b
2001  Guidewire          $3.9b
      SuccessFactors     $3.4b
      Bloom Energy         $3b
2002  LinkedIn            $27b
      SpaceX              $12b
      GoPro                $7b
2003  Tesla               $34b
      Splunk               $9b
      Tableau              $8b
2004  Facebook           $247b
      Palantir            $20b
2005  Workday             $15b
      Palo Alto Networks  $15b
2006  Twitter             $23b
      Lending Club       $5.7b
2007  Dropbox             $10b
      Fitbit               $7b
2008  Airbnb              $25b
      Cloudera           $4.1b
2009  Uber                $41b
      WhatsApp            $19b
      Square               $6b
2010  Pinterest           $11b
      WeWork              $10b
      Nest               $3.2b
2011  Wish                 $3b
      SoFi               $1.3b
2012  Instacart            $2b
2013  Zenefits           $4.5b

Observations and conclusions

The years when bubbles were bursting (2000-2002, 2008) were quite strong. So were the Dot-com bubble years (1997-1999) and the year before the real estate bubble popped (2007). Great companies are created every year. There’s no clear pattern of which years produce at least one $10b+ company and which years do not. Actually, the pattern is that almost every year produces a $10b+ company!

Because VC returns follow a power law, it doesn’t matter if there’s a bubble or not, it only matters that there are at least one or two huge companies created each year. Examples from the last two decades show that great companies are formed consistently, year after year, which means that even if we are near the peak of a bubble, founders should keep building companies and investors should keep investing in those companies.

** Valuations are: 1) market caps for publicly traded companies, 2) reported valuations for privately held companies, and 3) acquisition prices for acquired companies. All numbers are as of June 2015.