Is information technology behind industry concentration?

That is the topic of a new paper by James E. Bessen, and it appears the answer is yes:

Industry concentration has been rising in the US since 1980. Why? This paper explores the role of proprietary information technology systems (IT), which could increase industry concentration by raising the productivity of top firms relative to others. Using instrumental variable estimates, this paper finds that industry IT system use is strongly associated with the level and growth of industry concentration. The paper also finds that IT system use is associated with greater plant size, greater labor productivity, and greater operating margins for the top four firms in each industry compared to the rest. Successful IT systems appear to play a major role in the recent increases in industry concentration and in profit margins, moreso than a general decline in competition.

I expect further work in this area.


Is this driven by the fact that top firms are using IT better? Or is it simply that IT increases economies of scale (or reduces inefficiences related to scale) and thus increases the optimal firm size?

Also the inference of causation is potentially problematic for some of the measures. The most successful firms use IT more...but maybe they bought the IT systems because they were more successful and had more money than others.

Or is it simply that IT increases economies of scale (or reduces inefficiences related to scale) and thus increases the optimal firm size?

I would agree that IT increases the economies of scale as:
1) It is easier to measure individual workers
2) It is easier for everybody to get the data and control headcount. (In the 1980s and early 1990s the company leadership was experienced at IT reporting.)
3) It takes a lot of cash for good capital spending as IT capital is worthless to repo.

When all else fails with IT spending and usage, I simplify to the basic:

The marginal cost on IT usage is less than a penny and it is really cheap for new users.

Top firms likely write their own stuff. It is the knowledge about their processes expressed in an IT system. You need to be large enough to be able to afford and attract good developers who are capable of doing this stuff.

In a different industry, but same dynamic. Canadian retailers aren't large enough to write their own stuff, so are dependent on outside packages/consultants. Both Sobeys and Westfair have had IT deployment failures, Target Canada failed as a result of poor IT implementation.

It could also be that only large firms are able to survive the inevitable failures, or are able to implement the learn by failure structures within their large organizations.

I picked up on the adjetive "proprietary", that is IT gives ways in which a manufacturer can mix intellectual property (especially patents) into a machine that gives them monopoly-building tools.

Software patents are notoriously easy to get, so (for exmple) John Deere good add some software to their smart tractor and patent it so no one could add a similar feature to their own. Even without patents, they could make it so other machines that interoperated with it had to get a license etc. Or they could buy up a geographic database that was under copyright and prevent others from integrating with same.

All this encourages winner-take-all quasi-monopolies because they are most most valuable, and most available to whoever is already biggest in a particular field.

I don't know if you read Ben Thompson, but he is one of the most astute observers and writers about technology. He has invested many pixels developing "Aggregration Theory". Essentially an extension of Clayton Christensen's "Law of Conservation of Attractive Profits". Essentially, when there's zero marginal cost for adding customers, content and finding new customers, monopoly is the natural state. i.e. why Google, Facebook, etc. are so dominant.

I am not doing his writing justice, it's a theory you'd find interesting:

I feel like we have offered this mechanism in MR comments. Someone cribbing off our notes.

Of course, it's tech that has the greatest concentration, and tech is mainly about proprietary IT. Indeed, the tech business model, built as it is on advertising, thrives on scale. That's why Google and Facebook dominate digital advertising, capturing almost 70% of total revenues from digital advertising. If tech had taken a different path for its business model, I suspect that concentration would be far lower; indeed, the opportunities would be far greater But we have the tech business model that we have, not the business model that we might prefer.

What is "tech?"

If the Target checkout scans barcodes, and triggers smooth inventory replacement without human intervention, we just call that "retail."

That this can happen as easily with 1000 stores as 100 leads to increased concentration.

Indeed. It is a competitive advantage, a way to control costs, increase sales. The marginal cost isn't the advantage, it is the ability to actually do what retailers do consistently at low cost. The fixed costs of being able to do that, having a very good IT organization, attracting smart people needs to be amortized over a larger user base.

Well, I left my home and my one true love
East of the Ohio River
My father said we'd never wed
For I had me no gold nor silver

This is true for all disruptive technologies right? Automation for instance. But as tech evolves, it becomes cheaper and its power spreads out.

A more interesting conversation would be to analyze what kinds of IT are really at this early stage. For instance, online advertisement doesn't seem to be the case... we cannot just assume that because few companies provide a service that service provides power just for few players... (in the ad case, the biggest power is in advertisement itself, not the providing of that service)

In the case of retail it clearly amplifies economies of scale. The staff choosing and stocking fishing tackle can scale nationally for Walmart, but can't fall below 1 for a neighborhood hardware store.

Using retail as an example, you have companies nowadays that provide automation as part of a cloud solution. So retail companies don't necessarily need to know how to automate - they only pay a fee for that service. Of course there's still an economic barrier for retail companies that cannot afford services like that, but the number of companies using services like that is increasing exponentially.

Quickbooks and spreadsheets (small biz) always stink compared to well implemented SAP software (large companies). In essence, small companies, as a rule, are far more inefficient [sic], ineffective [sic] and bureaucratic [sic] than large companies, in this instance. However, like a broken German car, a broken, badly implemented SAP system is utterly punishing, even to the point of threatening to a businesses survival.

There's a saying abut SAP in tech: "you'd have to be a sap to buy their products"

The genreal impression I've gotten of them is that of byzantine, underengineered products made with antediluvian technologies (since they can't attract good programmers these days). Good code tends to come out of hedge funds / prop shops and some parts of Silicon Valley (though most of SV is just as bad as everything else). A good barometer is programmer salaries: if the average software engineer at the company is making less than 120k USD/yr, there's a problem.

IT, and how IT workers move, or choose not to move, between companies, also makes regional centralization stronger: You can trace all kinds of efficiency improvements and technologies across technology firms by just drawing lines between workers. Top firms might have done their best to stop poaching from each other through fun illegal agreements, but the main reason you can't start a big unicorn outside of SV is not that people out of there are less intelligent, but because all kinds of organizational knowledge transfers with developers migrating, and good luck hiring a group of people that have amassed all that social knowledge in Detroit.

I find this thesis doubtful at best. Did the large tech companies and China get an injection of American unicorn know how? Are NYC, Seattle and LA tech companies the product of organizational knowledge transfer?

Silicon Valley has a lot of talent, but developing software is quickly becoming commodified. The real trick is to get distribution, funding and a toehold in a comparative advantage that improves as you scale (network effects, economies of scale, regulatory arbitrage, etc.).

Related Question - to what degree do IT systems that improve customer total-cost-of-acquisition increase concentration?

That is, suppose both firm-A and firm-B have great internal IT systems. But firm-B's system allows customers to easily buy the right thing at the right time - so *customer purchasing productivity" and one imagines total customer productivity is improved by using firm-B's ordering system. Will that allow give firm-B a meaningful edge and growth towards concentration?

In last 15 years or so the amazon effect has been pretty obvious - has anybody tried to measure the value of time saved by people quickly finding what they want and having it delivered without hassle?

I know of some industrial systems with a similar effect - consider mcmaster-carr or MSC - both command a lot of market share in spite of NOT rock bottom prices, because for many users at many times, getting the right thing here soonest without hassle is what matters.

One advantage large corporations have is freedom from serious prosecution. See the new book "The Chickenshit Club" by Jesse Eisinger.

IT system use is associated with greater plant size, greater labor productivity

Comments for this post are closed