Why has productivity dispersion across firms gone up?

Yesterday Alex outlined the facts, which I take to be not in dispute.  Firms at the frontier have seen significant productivity gains, the others not so much.  Alex calls this a “lack of innovation diffusion” and considers whether IP law might be one cause.

My framing is somewhat different.  The result reminds me of the international trade literature on why so few firms export.  The notions of increasing returns to scale, and fixed costs to trade abroad, provide the beginnings of an answer.  In such a setting, let’s say the world has become more globalized, more IRS, and more based on learning curves, much of those trends being attributable to information technology.  In that case we would expect a growing bifurcation of firm productivity outcomes, just as we find a strong bifurcation of export outcomes, with a relatively small percentage of firms doing most of the international trade, or innovating, as the case may be.  The “only a small percentage of firms export” and the “only a small percentage of firms are on the productivity frontier” may sometimes even be the same way of describing the same basic fact.

The on the ground reality I observe is that the large, famous, exporting firms put together fantastic O-Ring teams of talent in a way the smaller, medium-size enterprises do not.  That is the relevant diffusion barrier, but of course there may be limits on that diffusion as well.  Eliminating barriers across firms is a good idea but not enough either.

And does the presence of relatively strict IP law subsidize such O-Ring teams, or limit their diffusion?  You can argue it either way.


I am a consultant who specializes in bringing new technologies to fortune 500 companies. Here are my 2 cents -
1)it is true that most firms do not "get it"
2) new technologies are largely human capital substitutes rather than complements (think of the impact on jobs of AI vs the typewriter)
3) many technologies require changes to customer behavior on top of changes to the production function (AOL still makes money from dial up) in order to incentivize firm change
4) don't underestimate the impact of regulation and rent seeking on firm behavior
5) the marginal cost of new technologies is much lower for frontier firms

I know this happens. I have seen it too. But consider the things Microsoft has been unable to do, with tons of brains and money and motivation. Their phones and their search were not so bad.

Something else is going on with second tier firms (MS as second tier now!) making things hard for them as well.

(Amazon's phone was also not so bad, but totally rejected by the market.)

There has definitely been a "first mover with the right arrow" and "internal corporate constraints can block success" issue in phones and search for MS. (Disclosure - I worked there for 20 years...)

The "first mover with the right arrow" meaning the first player with product correct for the market, with the right technology timing, rather than correct for some corporate goal or simply being the first to try it. This is why apple has had large success with largely technically mundane music players, phones, and tablets. In all of those cases, they were nowhere near the first mover. In at least two of those cases MS had useful products in the market years and in one case a decade before. But apple executed very well indeed, and focused on each particular market, rather than what it might have wanted internally.

By "internal corporate constraints" - at MS there is a deep mindset, which arises for obvious reasons, of "windows and office everywhere." And I certainly believed in that at various times. But now, I don't. I explicitly do NOT want to run the same apps on my phone as my PC, do NOT care or even want them to be from the same vendor. And so "windows and office everywhere" can be an obstacle in some markets.

I do not know, but suspect, that many large firms have similar issues.

Few firms export because most often supply doesn't magically meet demand a la Alibaba, but rather through personal connections, with extensive (and expensive) and deliberate actions to cultivate these relations. Language and cultural barriers get in the way of this bigtime.

I think this is rather different from the question of having just a few firms at the productivity frontier. (Although certainly these must be related - a local firm may often find some niche through local knowledge of how to market to this niche, but to export, you need massive advantages in product price, quality and/or brand recognition).

Imagine the scenario: you know you have a competitively priced product that is good quality, and that it competes with inferior and/or pricier substitutes in some particular market. You show up in a new market/country. You can`t read the local regulations and laws. You don`t know who the decision makers are. Every time you meet with people in pursuit of effective partners to expand the business, you offend people because you`re foreign and don`t understand their culture, or simply because the language barrier leads to miscommunication. You are limited to dealing with the minority of people who speak English (unless you have a very good translator who understands your business - very hard to find). The list goes on ... In short, even a significant productivity advantage isn`t good to get you far unless you have the deep pockets and huge patience required to cultivate the local connections.

Here`s for hoping that recent Alibaba forays into the American market will soon make all of this a matter of history.

Amazon scored two firsts, without much "IP", and by that I disregard "one click" explicitly.

Amazon simply became the first large successful online seller, with books, and then ratcheted that from books to general merchandise using their early-formed cumulative advantage. Pretty much from day one (perhaps I should say "day two") it was a low technical hurdle to create "an Amazon," but very difficult to get traffic for it. The "technology" is seen to "lack diffusion."

As a side effect of their economies of scale they created the first great cloud serving infrastructure, and they now tower over any other entrant. It might be a low-ish technical hurdle to create "a cloud," but hard to build a 1 million server cloud. Few people have those deep pockets. So few others can compete. The "technology" is seen to "lack diffusion."

A seeming contradiction of the internet, where the barriers are so low, yet so large.

a) it is difficult to create an IaaS service which is actually stable. It requires a tremendous amount of engineering on the software side. Further, on the HW side driving down costs requires enormous engineering. It isn't simply economies of scale, it is just buy a bunch of servers.

b) it is so much more than just IaaS, AWS has so many services that are used by many of their clients. S3, EBS, RDS, EMR, the list goes on and on and each of these services requires vast amounts of engineering.

There is a reason why Dell, HP, Cisco, or some other random hardware company some cash doesn't have a cloud service. Cloud is a software business and requires mind boggling amounts of engineering.

The people you name, Dell, HP, Cisco cannot create a cloud without competing with their customers, burning their bridges.

Others have built clouds out of Dell, HP, Cisco tech and open software stacks.

But as I say only a handful can field a million servers and that gives them huge cumulative advantage, including the software tech that handful developed along the way.

I am certainly not pretending Amazon, or Google, or Microsoft learned nothing as they grew.

Look at the very large difference in productivity gains between frontier and non-frontier firms in the curves Alex is showing. If there are frontier and non-frontier firms in a product segment, the implied difference in profitability is enormous. I would say that the the non-frontier firms are, by now, dead.

So, do we have segments with all frontier firms (maybe only one), and segments with all non-frontier, and none with a mix? Or how do you interpret this?

I think IP must severely hurt O-Ring style production, since it makes many patent holders essential links in the production chain. Even if the patent holders happen to be of a high quality, their monopoly right eliminates the possibility of low and medium quality versions, and therefore hurts optimal matching. Consider that if patent rights flow to the highest value producers, that only allows matching at the top end of the distribution while starving all other firms. So an o-ring story with IP is still consistent with large variation in firm productivity due to weak innovation diffusion.

I buy the high fixed costs of intangible goods and the rest of the logic follows from that. I don't see the need to introduce the O-ring metaphor. That is an over-rated paper if there was one. It simply isn't needed to explain these facts.

On a related note, renting platforms like AWS and others from google etc. allow cheap access to large scale computing power and the continuing explosion in open source code allows for the free availability and use of sophisticated software as a base for further applications. These and other new technologies will allow small groups of people to benefit from technologies that were only available to large firms, even just a decade ago. As human capital investments respond to incentives we will see even more rapid technological progress in the world of "bits" as Peter Thiel likes to categorize these things. The world of atoms still requires old fashioned scale economies - though sophisticated horizontal and vertical contracting may help a bit.

That cheap computing has coincided with concentrated investment in the Unicorns. And those Unicorn companies are the ones, the few, with astronomical worker productivity.

Again, Uber and AirBnB are not technically difficult, but their valuations presume low chance of competition.

The market expects rapid creation of cumulative advantage.

The 'O Ring' notion is bullshit.

& no reference to Michael Porter's 'Cluster' model of innovation of twenty five plus years ago? ['Competitive Strategy' & 'The Competitive Advantage of Nations']

The examples you mentioned involve network externalities, not an accumulation of specific highly trained human capital, or as you put it : "astronomical worker productivity" . In addition, at least for Uber (though perhaps also for Air B&B), the real competitive advantage remains moxie and a relative willingness to be unencumbered by ethical considerations - without which it would be have been difficult to break existing monopolies supported by rents and licensing (and the associated relaxed ethics).

i say the same comment of Bryan Willians
"I do not know, but suspect, that many large firms have similar issues"

Cumulative advantage is about network effects in large part but, and this is important, it then shows up as worker productivity.

AirBnB makes much more per worker than companies operating under previous models.

I see that the logic can get convoluted. There are other ways to look at this, and I might be wrong. What I meant was that the basis for profitability ("super-normal profits" might be an old fashioned description) - the entry barrier that prevents real competition - comes from the demand side (network externalities) and not from the the production side (which for example, might be something genuinely innovative and impossible to replicate by another team of highly skilled workers).
This is consistent with Bryan Willman's “first mover with the right arrow” description.

e.g., once everyone is on Facebook, how can someone replace them? This will prevent competition. Once sellers are all on Amazon and eBay, how can someone replace them? This will prevent competition.

However, I think they will not abuse their position "excessively", because technologically speaking it is very easy to do what they have done.

First sentence : yes. This is what network externalities are : https://en.wikipedia.org/wiki/Network_effect.

Second sentence : does not follow from the first. Also how do you define "abuse" and "excessively" ? Quite tricky, especially in the technology world.

"abuse" as "excessive" monopolistic pricing.

Of course we expect them to enjoy some degree of pricing power, but "excessive" would be somewhat subjective. To define "excessive" I would consider the following situation. They raise their fees and pricing on a lot of things. This attracts competition who observes their profits. And then the lower prices dramatically to kill the competition. If that sequence of events passes, then it is "excessive" and they have also broken the law.

I'm not sure that it would make sense to appeal to a strictly formal definition of this. "Excessive" would be whatever leads to a wide public outcry.

This could be describing the elephant, all true, but different perspectives.

FWIW, I think the most interesting thing is, as I say, the expectation (often confirmed in the market) that cumulative advantage can be easily created. How much that depends on the arrow, and how much that depends on moving first, might vary a bit, but I think we agree that technology is pretty broadly available, as a backdrop.

It seems obvious that the differentials in productivity growth are between different fields and not between head-to-head competitors in the same fields. If it were a case of some firms "getting it" and others not, there would be "not getting its" failing all over the place.

Mostly this reflects that in some business fields technology is advancing rapidly and in others it's not. Also keep in mind that intra-group transfer pricing typically suppresses apparent productivity growth in core units and inflates it in business-service units.

More evidence on this here http://econweb.umd.edu/~haltiwan/DHJM_6_2_2014.pdf

And i should say, not only is this an active research area, but it is a very old literature. Try Lambson (1991) for a start, http://www.sciencedirect.com/science/article/pii/S0167718705800013

Never in the history of economics has there been a less puzzling observation, or question, as this.

It's the whole point of doing business.

Not to mention that there's whole disciplines dedicated specifically to this, which apparently the two economists running this blog have never read of or heard of, but instead call it all just "dark matter".

PS: I would recommend you take a look at the thousands of papers on innovation in various disciplines, from economics to finance to management. But barring that, I'd recommend you start off with your own "Austrian" roots at Schumpeter and Hayek. Instead of looking for boogeymen and government regulations and market distortions, first take a look if this is precisely the intended outcome of competition.

If there's whole disciplines dedicated to this, keywords would be useful

Maybe the lack of diffusion is based on geographical factors. The various firms are scattered hither and yon. The expertise in the "new" productivity instrumentality is concentrated in Silicon Valley and to a lesser extent other major intellectual centers.

noxjpfaqetzFObgLhT 4346

'The on the ground reality I observe is that the large, famous, exporting firms put together fantastic O-Ring teams of talent in a way the smaller, medium-size enterprises do not.'

The German noun that refutes this entire framework as somehow being universal or particularly useful descriptively is 'Mittelstand.'

Comments for this post are closed