1/9/15

Are There Too Many Startups? Part 1: Definitions and Data


Like always, there is something missing here. Hint, it starts with F, G and W. 
It’s hard to say when exactly the “startup craze” began. Of course there was the dot-com bubble, but that burst, and it seemed for a time that early-stage tech companies had lost their fizz, at least compared to mortgage backed securities. Perhaps it began anew when Apple launched the first iPhone in 2007, or when Facebook launched for high school in September of 2005. It was certainly fully underway by 2010, when “The Social Network” came out,  and by 2011, many were calling it another bubble.

Two years later, startups seem, if anything, bigger. All the kids want to move to California. Every other news article is about Uber. There is a parody sitcom. But is this explosion in popularity a dangerous display of irrational exuberance, a shift that should be celebrated and encouraged, or a false perception? Are there too many startups, or are there not enough? Many people believe in entrepreneurship as a driver of growth. This belief has the power to drive policy, philanthropy, and for-profit investment. But it could be misplaced. I believe that where there is smoke there is usually fire; the truth, of course, is that nobody knows.

Google trends chart for "Uber." Where there's smoke there's fire? 

As an alumni of Venture for America, this is a topic I am personally invested in (and am perhaps unqualified to study for that reason). From 2012-2014 I was part of an organization whose stated mission is to recruit graduates away from careers in finance, law, or consulting, and towards entrepreneurship; an organization that wants to create as many new entrepreneurs and startups, as possible. These posts are my attempt to corral the evidence, empirical and theoretical, into one place, and to develop a framework on how to approach the question.

Finding "Startups" in the Data 


First and foremost, a definition. What is astartup?” Wikipedia defines it vaguely: “a company, a partnership or temporary organization designed to search for a repeatable and scalable business model.” 

Some people, often unconsciously, define startups by INDUSTRY. Considering the nature of this startup craze, this definition may not be as misplaced as it seems. If what we are really excited about is economic potential that new technologies, like mobile and more advanced data science, have unlocked, than looking at data on technology firms may give us more insight than any other measure. But most of the mythology around startups these days is industry agnostic, and drawing the line between technology companies and companies that use technology to solve old problems (like Uber) is often difficult. 

Some people define a startups by firm SIZE, but size conflates the innovative, high-growth darlings of popular culture with an entirely different organism, the Mom & Pop shop. Our national nostalgia for Mom & Pop shops is clearly indicated by the name, and also by their frequent appearance in political speeches (usually via the less precise moniker, “small business”). 

Young small businesses grow fast. Old small businesses shrink. 

But unfortunately, according to economic theory and empirical evidence, Mom & Pop shops aren't all great. They are the opposite of Adam Smith’s pin factory, his famous illustration of division of labor, and empirical evidence confirms that large firms tend to be more productive than small firmsThey also don’t create as many jobsand employees tend to be more vulnerable to economic fluctuations and, on average, don't earn as much. We may still want to save the Mom & Pop store, but if we do, it should be for reasons other than economic growth. 


Small businesses on average pay less. http://www.kc.frb.org/PUBLICAT/ECONREV/PDF/2q07edmi.pdf
Many economic journal articles written in the last 20 years have used SELF-EMPLOYMENT as a proxy for entrepreneurship, mainly because it’s a variable in several popular, publicly maintained datasets on individuals. I find the use of self-employment in articles about “entrepreneurship” to be especially problematic, because freelance writers, artists and developers, independent consultants and lawyers, and most agricultural workers are not entrepreneurs. Better studies differentiate the “incorporated” self-employed from the “unincorporated” self-employed, but as this metric still captures most of the Mamas and the Papas, it doesn’t really help us answer the question.  


The incorporated self-employed make up a less than a third of the self-employed. http://www.bls.gov/opub/mlr/2010/09/art2full.pdf
The best metric for studying startups is probably firm AGE, as University of Maryland economists Ryan Decker, John Haltiwanger, Ron Jarmin, and Javier Miranda argue. The cutoff is somewhat arbitrary; they call a firm a startup if it is less than 5 years old, but 3 or 10 might do equally as well. Age still doesn't provide a crystal clear story, because it captures at least two very different types of firms: the Silicon Valley troupe, but also new Mom & Pop shops, new construction companies, nail salons, or cronut stops. Do we care about both? 

Another metric for studying the startups could be VENTURE FUNDED firms. Venture-funded firms do hog most of the media coverage of startups, but they also represent a considerably smaller slice of the economy than any other metric on this list, almost to small for worthwhile analysis. The data is also harder to come by, and often incomplete, or (for me) expensive. One point in this metrics’s favor: after the financial crisis, most banks reduced lending to small businesses declined dramatically, and so more entrepreneurs seeking funding in the last 5-10 years have had to turn to equity markets. However many entrepreneurs don't raise venture funding at all, instead relying on capital from friends & family or early revenues to support growth. 

So far we’ve listed three bosh metrics, size, industry, and self-employment, and two alright metrics, age and venture-funded-ness. Of course in economics there are two ways to measure everything, units and money. When looking at the question “Are There Two Many Startups,” both will be interesting.  

How Many Startups Are There, Actually? 


First let’s look at age. According to the article by Decker et al, the number of new firms have been declining steadily for about 30 years. What what? 

The decline in the rate of job creation and destruction has been noted my multiple other sources over the last ten years, so it's pretty unlikely the data is wrong. But how can there be so much smoke and so little fire? Is this another obvious thing I just didn't see, or are we using the wrong metric?

The venture-funded metric tells a much more familiar story. According to Mattermark, a private company that compiles data from a number of venture capital sources, “Q4 2014 Startup Investment is at the Highest Point in 10 Years": 



Mattermark’s graph certainly makes startup craze 2.0 look suspicious. But when you zoom out a little, you can see that the recent uptick is dwarfed by the dot-com bubble of the late 1990s. This is good news for the entrepreneurship optimists.
National Venture Capital Association Yearbook (NVCA)
When you measure in firms instead of money, the story doesn't look too different. 

Why so different? Here are a few theories. 

  1. It's possible that the startup craze of popular media is indicative not of an increase in the number of startups but of the number of quality startups. Or, of course, that the startup craze is in fact, a bubble, and investors are pumping money into a declining number of new firms without any solid economic justification. 
    1. Age could be capturing a decline in the incorporation of Mom & Pops shops, not in the innovative firms we care about. This is what Noah Smith suggests, and the rise of big box retailers combined with the ongoing shift away from manufacturing towards services and retail is a possible explanation offered by Decker et al in an earlier paper. The age story could be just the next step in the structure of employment, and not particularly indicative of a change in in the rate of innovation. Decker et al are working on an answer to this now, I think. 
      Plausible? Sure. 
    2. Ben Casselmen at fivethirtyeight points out that Brookings points out that new firms and population growth are highly correlated, and population growth has been declining steadily for some time. (There was, however, an echo-boom spike around 1990 - will my generation take back the night?) Much of the entrepreneurship literature stresses the difference between "opportunity" and "necessity" entrepreneurs; entrepreneurs who see an innovation market opportunity vs. entrepreneurs who just need to make a buck. You can guess which group is ore likely to be venture-funded. Would it make sense that the decline in entrepreneurship related to population growth is mainly of necessity variety? (In other words, "Damn, there are more people than there used to be... we gotta start a company!" 
    3. One of the clearest trends in the Mattermark, NVCA & other data is the increase in the average age of venture-funded firms. (See, for instance, the increase in follow-on relative to first round investments above). Is it reasonable to conclude that the average age of startups has increased for some structural reason? Maybe as technology has gotten more complicated, companies need more time to create a market-ready product, or in general the whole life-cycle of industry has gotten longer. Eh? Eh? 
    If you can tell, I’ve been rabidly holding on to my qualitative perception that there are more startups than there used to be. But in reality, neither the number nor the trend of startups definitely answers the question. To understand why, we’ll have to turn to theory. Stay tuned for part 2! 

    Bye bye now 

    No comments :

    Post a Comment