Annotated: “Want More Startups? Build a Better Safety Net”

Wherein I annotate things.

Today, responding to Noah Smith’s article, “Want More Startups?  Build a better Safety Net”:

Back in 2012, Daron Acemoglu — an economist I follow and greatly respect — wrote a paper along with James Robinson and Thierry Verdier claiming to explain why Scandinavian countries are (supposedly) less innovative than the U.S. Acemoglu et al. theorized that Scandinavia embraces “cuddly capitalism” — a strong safety net that prevents failure — while the U.S. goes in for “cutthroat capitalism.” The do-or-die nature of the American system, they said, causes people to try a lot harder at innovation than their European counterparts.

Bloggers were quick to point out that the paper’s entire premise was probably wrong. Scandinavia isn’t, in fact, less innovative than the U.S. Acemoglu et al. used patents as their measure of innovation, but the volume of patents is more about intellectual property law than entrepreneurial effort. Broader measures of innovation show that the U.S. and Scandinavian countries are about equal. Acemoglu et al. had constructed an elaborate theory to explain something that probably never existed.

Noah cites a Matt Yglesias post about how the innovation literature is (wrongly) obsessed with using patents to measure innovation.  And hey, it turns out we’re still doing that all the goddamn time (I still frequent academic econ conferences on innovation)!

I know that most of the post is not about this, but it’s still a central, massive issue in the econ literature.  The reason why is understandable: the data are very easily accessible, stretch back 30+ years in a clean format, and a lot of work has been done already such that matching into, say, private firm datasets is easy.  Moreover the alternative here isn’t really clear — what else would we use?

But for all sorts of reasons [note to me: which requires another post entirely], this makes for some really, really limited research in terms of useful descriptions and predictions about the world.  I guess the short version is that we desperately need some new ideas (and data!) for measuring innovation.  Hopefully I will get to work on this soon…

So what was wrong with the theory? In Acemoglu et al.’s model universe, the thing that determines entrepreneurial success isn’t how smart you are, how much risk you can take, how good the business climate is, or how much demand there is in the economy — it’s how hard you try.

Effort-based explanations are relatively common in economics and econ punditry, especially among those who lean toward a free-market ideology. I occasionally hear people say that southern Europe is economically worse off than northern Europe because the former has a weak work ethic. Others claim that welfare makes people lazy. And some economic theories even attribute recessions to a great vacation — a sudden decrease in willingness to work. In a recent Bloomberg View debate, my colleague Tyler Cowen suggested that lower employment levels since the Great Recession may be caused by more men wanting to stay home and play video games.

Generally I like having Tyler Cowen around but… lol.

When it comes to getting a job, the evidence for the effort hypothesis is mixed. Most studies show that the creation of cash-transfer programs doesn’t discourage work, while there is evidence that the U.S. welfare reform in the 1990s did force some people back into the workforce.

This strikes me as kind of tangential to the point.  We care about social programs’ effects on entrepreneurship / startup rates / composition of startups (e.g., we might expect more growth startups with higher risk insurance, vs. a 2 person “consulting” firm with no offices), which is a separate issue from discouraging work-as-an-employee.

Also for more on the discouraging work question, via Nick Bunker over at WCEG, there is some evidence for cash-transfer programs mildly discouraging work, so yep all agreed the literature does not seem to be settled on this.

But when it comes to entrepreneurship, there’s another factor that’s probably a lot more important than effort. It’s risk.


For a prospective entrepreneur, the choice usually isn’t between starting a business and playing video games — it’s between starting a business and working for someone else. The difference in effort between those two career paths probably isn’t that big. But entrepreneurship is much, much riskier than holding a job. In general, you’re a lot more likely to see your business fail than to be fired.

Well now hey “the difference in effort between those two career paths probably isn’t that big,” that just cannot be right.  But yes, much riskier, for sure.

The risk theory of entrepreneurship says that when people already have a lot of risk, they’re less likely to take on more. This is just because most people are risk-averse — if they’re under threat of losing their job or paying a huge medical bill, they’re probably going to be less willing to gamble all of their savings on starting a company.

Generally, I think this is right.  But we can probably even model this more simplistically by rolling risk into the general “costs” of entrepreneurship.  More social insurance reduces the (risk-adjusted) cost of starting a startup.  Fundamentally this is identical, but I think useful to think about because it might get us thinking about other ways to reduce the costs of entrepreneurship beyond indirect (and probably unintentional, French aside) risk adjustments via the social programs channel.  And/or it helps us think about why buzzwords like “business climate” (e.g., # of days to get licenses) may not matter very much, or may matter by degrees.

It can also get us thinking about the broader reasons behind declining business dynamism, but more on that momentarily…

So if the risk theory is right, a stronger safety net should lead to more entrepreneurial activity, not less. Whatever negative effect public assistance has on effort will be more than canceled out by the greater risk-taking capacity of the poor and unemployed.

Wait wha- hold on there, that’s an implication too far.  I mean sure I’d love for it to be true, but “more than canceled out” is a potentially huge claim, and I don’t really think the cited studies establish that.

That’s what the evidence seems to indicate. A 2004 study showed that food stamps make people more likely to open a business. Another study in 2012 demonstrated that giving extended unemployment benefits also makes people more likely to start companies.

Just want to take time out to say, if you (dear reader) have not read the Schoar paper (second one cited above, from 2012), it is magnificent, go do that.

Now we have yet another piece of evidence. Economists Joshua Gottlieb, Richard Townsend and Ting Xu studied the effects of maternity leave in Canada. They found that women who had the guaranteed option to return to their jobs after maternity leave were more likely to start businesses. The reduction in risk from having the fallback option made entrepreneurs more willing to take the plunge. That’s the opposite of what an effort-based model would predict.

Townsend moved to UCSD?  Anyway.  I need to take some time to read through the paper entirely, but this speaks to my point earlier: “The results are driven by more educated entrepreneurs, starting firms that survive at least five years and hire paid employees, in industries where experimentation is more valuable.”  If I’m interpreting them right: it’s not only that startup rates generally increase, which is nice, but in fact the composition shifts more toward growth entrepreneurship or at least firms that survive (no mean feat).

So the data is piling up — at the margin, risk is a lot more important than effort in determining who starts a business. With the U.S. suffering declining entrepreneurship rates, this has big implications for policy. Instead of making American capitalism less cuddly, government should focus on finding ways to limit personal risk. That could include single-payer health care, expanded parental leave and other social safety-net programs. It could also include nudging households to take on less debt, since debt creates risk. But whatever the solution is, making the economy more cutthroat seems likely to hurt the goal of boosting business dynamism.

So, on declining entrepreneurship rates generally.  Obviously see Decker, Miranda, and Haltiwanger’s ongoing work on the issue.  I believe the published viewpoint is that up to 2000 it’s about the rise of big-box stores, and post-2000 there’s no great explanation yet but Javi’s latest (unpublished) viewpoint, as far as I remember, was something about the rising costs of starting a competitive business over time.

Which is to say that I don’t think the social safety net is related to the broader decline up to this point (not that Noah makes that claim, mind).  So, while I’m wary that enacting a stronger social safety net would completely reverse the secular decline in business dynamism, I do agree with Noah that the evidence is piling up that it would have some positive effect on startup rates (and maybe even composition), and that’s a very nice added benefit, and one that almost never comes up in policy discussions on these things (although to be fair I recall a bit of a blog debate on it re: ACA, at time of implementation).

But, returning to the point about risks vs. costs above, I think that if we want to encourage entrepreneurship (and by extension, the innovation we hope comes out of it) via policy, we should be doing it both more directly and more intentionally, rather than just counting on spillovers from other programs.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s