Bryan Caplan  

The Decline of Creative Destruction

McKenzie on Driving vs. Walkin... We Won!...
Though I've been a harsh critic of Tyler's Great Stagnation thesis, I was struck by the following figure in Edmund Phelps' Mass Flourishing:


At first glance, this confirms a quarter-century of steadily declining creative destruction - falling job creation and job destruction.  On closer look, though, there was little trend until the small recession of the early 2000s.  Since then, however, creative destruction has relentlessly fallen. 

Striking fact: The rate of job destruction during the Great Recession used to be perfectly normal!  We experienced it as a calamity because job creation not only kept falling, but dipped below expectations.

Anyone seen this diagram over a longer time horizon?  For other countries?  Inquiring minds want to know.

Comments and Sharing

COMMENTS (22 to date)
Pajser writes:

I'd expect such curves on approximation of the free market. Economy of scale → concentration of the capital → more work is redistributed without change of the employer.

david writes:

I feel tempted to suggest that job creation Granger-causes job destruction in that graph, particularly outside recession periods, but it's hard to tell without actual numbers.

Gabriel writes:

This is not surprising...

Most jobs are created within new companies, see this chart (here) from an article on pando daily (here). If the company is older than 11 years, odds are that they are on average shedding more jobs than they are creating.

New business creation, even with all the tech startup craze of the last few years seems to be at a 30-year low (see here), and has been falling over the last decade... which seems to correlate with the graph showing less creative destruction and creation, since there are less businesses started (and destroyed) every year.

Finster writes:

The answer of course is not to look at a closed national economy, but rather Michael Pettis' "Chimerica" economy.
Job creation has followed/accompanied job destruction, but the overseas container and capital mobility mean they have not been created within the delineations of the US economy you use in the statistic.

Ryan writes:


This downward trend goes back at least to the late 1970s. Prior that that time, we only have data for manufacturing--but in mfg the trend goes back even further. This is a long-run trend, though it does accelerate after the late 1990s.

Here is a paper documenting the trend and showing that it probably cannot be explained by composition effects (but, as Gabriel notes, the aging of the firm distribution explains some of it--see here and here).

Here is my comment on Marginal Revolution from a few months ago, discussing some of the composition stuff briefly. Here is a discussion of trends in establishment entry and exit, with big implications for understanding the Great Recession.

Finally, here is a summary of a recent paper that looks at the JC/JD trends as well as worker flows data (where a similar trend is observed).

What we know is that (a) this decline is observable in several dynamism series (job flows, worker flows, firm dynamics, establishment dynamics), (b) some composition effects--specifically firm age--can explain some of it (but we can't explain what's causing the age stuff); other composition effects (like industry) work in the "wrong" direction (ie, make the puzzle more puzzling), and (c) we don't really know what's going on.

Since we don't know what's causing it, we don't know if it's good or bad. We typically think of creative destruction as good, but how much of it do we need? Is the current amount too low, or was the previous amount too high? After all, exit and reallocation and job switching are costly. Are we closer now to the right amount, while previously the US economy was just a Rube Goldberg machine, doing a bunch of unnecessary churning? Or does the current level of dynamism reflect policy or technology barriers to efficient allocation?

BP writes:

(1) Aging of the workforce. Older workers hold their jobs longer. Could be caused by rising unemployment among the young.

(2) The modern knowledge worker can more easily redefine his job within the same firm without explicit creation or destruction. Knowledge work could also be increasing as a share of total work.

Does job creation/destruction matter as much as real gdp growth? Which economy is better?
Economy A: High creative destructive, low real gdp growth per capita
Economy B: Low creative destruction, high real gdp growth per capita

Ryan writes:

BP--Great observations.

(1) Hyatt and Spletzer (2013) find that the aging workforce has some explanatory power, but not enough.

(2) Interesting point. We should think of a way to measure this. We can get at this question somewhat indirectly by looking at industry and occupation; my understanding is that those are also insufficient explanations.

Your final point is the key. Without knowing more, we can't assume that the decline is a bad thing (or a good thing).

tew writes:


I'm not sure your first point (1) is valid, because I don't think it fits the definition of job creation and job destruction. I think job destruction is defined as the elimination of a position at a firm (usually accompanied by a layoff unless the same firm creates a new job for the person). Similarly job creation would be the addition of headcount at a new or existing firm. So less frequent job hopping would not impact this.

Foobarista writes:

I wonder how temp jobs and underground (off-the-books) jobs figure into his number? Both have boomed in recent times.

Jack writes:

Hey Bryan,
O/T This got buried in a previous comment section, but I'm curious-

Do you believe Israel should open their borders?


Andrew_FL writes:

It strikes me that the graph begins in 1990 and I can't help but wonder if this amounts to cherry picking. Might it really be the case that both were unusually high by historic norms in the 90's?

I guess this element of the argument is that higher turnover is better than lower turnover-or that the higher turnover in the past was better than the lower turnover in more recent years (ie, since ca. 2000). But I don't think that is obviously true. The important element of capitalism is *that* there is creative destruction, not the *amount* of creative destruction. It would be more distressing if the two curves *didn't* move together than if they were both flat or both going up, or what actually happened.

Ryan writes:

Andrew_FL--the chart begins around 1990 because it's based on BLS BED data, which only goes back to around 1990 for these items. See my comment above--we can see this trend going back decades earlier in other data sources (for example, in the BDS).

Andrew_FL writes:

@Ryan- Thanks. So that means there are at least arguments that this trend probably did not just begin recently, but is a long term economic trend. As far as I can tell, you do agree in principle with what I said about this not *necessarily* being a bad thing. I would agree that we should want to know what is causing the trend to better determine whether it is something we should worry about.

One question I would ask is whether this is something unique to the US or if it might be something that has happened or is happening in other countries, which might give us a clue as to whether we should simply expect this kind of thing to occur in an increasingly large economy.

Drea writes:


Another factor I'd love to know about: how do people who are self employed, freelancers, or have a business of one show up in these numbers?

I've heard in other venues that the distribution of small businesses is shrinking at the middle, but growing at the 1 person size. This ties into the anecdotal "freelance nation" trend. If "jobs" aren't being created inside the firm, but instead more people are working independently, that might show up in the decline in both JD & JC.

Enial Cattesi writes:

As far as I remember, Cowen's theory is that we reached the great stagnation because of some limits in human mental capabilities.

Cowen at one point said that increased state power was accompanied with increased freedom, so it is explainable why he would think that the great stagnation would not be caused by a wide and thick web of regulations, but by us reaching our mental limits.

The theory sounded silly then and it didn't get better with age.

Andrew_FL writes:

@Enial Cattesi- Did Cowen really say "increased state power was accompanied with increased freedom"? One certainly hopes not, since that is a contradiction in terms.

Daniel Kuehn writes:

Older data is usually restricted to manufacturing - see Haltiwanger's website. The business dynamics program at Census might go back further than this, I forget when they start.

NBER has a series from the interwar period but I believe it's just worker flows, not job flows that you're interested in here.

Enial Cattesi writes:


Those developments have brought us much greater wealth and much greater liberty, at least in the positive sense of greater life opportunities. They’ve also brought much bigger government. The more wealth we have, the more government we can afford. Furthermore, the better government operates, the more government people will demand. That is the fundamental paradox of libertarianism. Many initial victories bring later defeats.

From here:

Also (this is interesting):

The old formulas were “big government is bad” and “liberty is good,” but these are not exactly equal in their implications. The second motto — “liberty is good” — is the more important. And the older story of “big government crushes liberty” is being superseded by “advances in liberty bring bigger government.”

And it goes on and on and on.

Andrew_FL writes:

I suppose you *could* make the case that a government that performs essential functions does need to remain proportional to the population or economy, but he isn't speaking in the hypothetical, but of what has actually happened, where government has grow disproportionately.

I don't think it is remotely true that has lead to an expansion of liberty or even been associated with one.

[edited with permission of commenter. --Econlib Ed.]

teapartydoc writes:

Feudal societies have great job stability and few jobs with little or no creative destruction. That is the direction in which we are headed.

Bear writes:


Your question assumes that it's even possible to have high per capita GDP growth with low creative destruction.

Can you cite an example of when/where this has happens or is happening (sustained, not just a short boom)?

Just off the cuff, the notion of a high-growth/low-risk/relatively static economy strikes me as too similar to the radio ads for some "investment" that promise "all the upside growth of the stock market, with NONE of the downside risk!"

Gary Hemminger writes:

This graph doesn't surprise me at all. To me this is expected of economies that are increasingly regulated. Regulation makes it difficult to shed jobs and also create them.

Comments for this entry have been closed
Return to top