Software Is Not Capital if You’re Not a Software Company
The Economist trumpets Thomas Piketty’s Capital and his theory, r > g, has had its first serious rebuttal, glowing like a proud parent over graduate student Matthew Rognlie’s work.
Note this bit:
Mr Rognlie mounts three main criticisms of these arguments. First, he argues that the rate of return from capital probably declines over the long run, rather than remaining high as Mr Piketty suggests, due to the law of diminishing marginal returns. Modern forms of capital, such as software, depreciate faster in value than equipment did in the past: a giant metal press might have a working life of decades while a new piece of database-management software will be obsolete in a few years at most. This means that although gross returns from wealth may well be rising, they may not necessarily be growing in net terms, since a large share of the gains that flow to owners of capital must be reinvested.
Emphasis mine.
Most commercial software used by corporations, including the example of database-management software, is licensed. Users are licensees, not owners.
Software doesn’t necessarily obsolesce, either. I’ve worked for businesses using software that was as much as twenty years old. Small businesses, in particular, can continue to run well on old accounting software, provided they don’t need highly granular reporting.
What does become obsolete is the hardware. If software no longer runs on an older system, or if it is no longer serviced by the licensor (ex: Windows XP), the licensee has simply reached the limit of the license.
This includes upgrades by software manufacturers for reasons of security improvements: if users don’t upgrade for improved security, they’re outside the limits of the license.
The only entities that might be able to claim software is capital are software companies. This might not even be the case if capital is limited to the licenses they’ve granted and claimed as assets — any accountant, tax attorney or IP attorney want to respond to this?
The confusion about software’s nature probably lies in our accounting and tax systems, which may treat software as an amortizable intangible asset. (Feel free to correct me in comments as I am not an accountant, nor a tax preparer, nor a tax attorney.)
But most commercial software remains a licensed product.
Companies are also moving toward “software as a service” (SaaS), provided a license to access software on software providers’ systems. Microsoft’s Office 365, Google Apps, Salesforce.com are examples of SaaS. There are even further reductions in companies’ need for investment in hardware when subscribing to “infrastructure as a service” and “platform as a service,” like IBM, Amazon, and other technology companies offer.
These are contracted services — definitely not rapidly depreciating capital assets.
What exactly does Rognlie mean by “modern forms of capital” when his understanding of software is flawed?
I haven’t looked deeply at the rest of the arguments Rognlie offered as a rebuttal to Piketty’s theory. This bit checked me short, giving me concerns about his remaining points addressing returns on wealth, and on distribution of net capital income.
[UPDATE: Do read Ed Walker’s comment about this piece in The Economist.]
Rognlie avoids calling humans the actual source of capital – the real means of production. Software in itself is little better than a leased screwdriver; it cannot create value without a human’s intellect applied.
It’s more than a leased screwdriver to the corporations producing software, but the production of the software itself relies on human intellect.
IMO, we really need to examine not “modern forms of capital,” but the definition of capital, as Ed Walker (massacio) has already been doing here. Humans have been treated as fungibles until now, but if companies’ business models rely on human intellect, is it merely a commodified input, or another form of capital?
And what happens to our culture if we begin to see every human as a source of capital?
This article is absurd. First, Piketty addresses the depreciation idea specifically. See p. 178, where he explains that savings are net of depreciation. Savings are crucial to his entire thesis, because the richest have the greatest savings, which explains why their wealth increases so much faster than that of everyone else.
Second, depreciation is relevant to owned assets, but not to leased assets, as Rayne correctly notes. The entire cost of leasing software is expensed as incurred (or paid, in the case of cash basis taxpayers.)
Third, the purchase of software is heavily tax-advantaged. You write it off much more quickly than it actually depreciates. This is true even though for the price of a subscription, you get plenty more use from the software.
Fourth, this guy doesn’t seem to grasp the fact that intellectual property rights have been increasing as a result of legislative changes bought and paid for by corporations. This is in perfect accord with Piketty’s view that capital is increasing because of structural changes in the way society works. The point is that protecting IP gives corporations the ability to extract money from the rest of us in new ways. That cuts completely against Rognlie’s point that the returns to capital should decline with competition. In fact, there is every reason to assume that the added power of wealth will increase the ability of the filthy rich to extract money from the rest of us.
Rognlie makes a big point about land values and their increase, and claims that all the money is going to landlords. He makes it sound like the rentiers are a bunch of doctors and lawyers buying up the land. That is false. The gigantic owners of property are insurance companies, pension plans (TIAA-CREF is a major owner of property), investor-owned REITS, oil companies and of course the gigantic developer corporations and the gigantic real estate companies, and hotel/motel companies.
Land doesn’t depreciate. The buildings do, but at a slower rate than allowable depreciation in almost all cases. The price of land is again a function of the use of zoning laws to limit growth, and can easily be reversed by legislative changes.
Noah Smith commented on this paper in Bloomberg, arguing that it implies the need for a higher land value tax. Piketty calls for legislative changes to limit the growth of capital, including an annual tax on capital wealth. That, of course, is the exact function of a land value tax. I don’t see a general call for a wealth tax in this Economist summary, though.
There are plenty of questions about Capital that make sense, including the extent to which returns to capital will remain high. This shoddy discussion doesn’t get to any of them. I won’t be reading the paper itself unless someone tells me it makes better sense than either the Economist or the Bloomberg piece shows.
Your impression is closer than Rognile’s. He has bought in to the marketing fluff of the software industry. Much of our larger institutional and corporate software is thirty or forty years old and is only subject to hardware (platform) obsolescence. If there is a compatible compiler, the software can be “ported” to the new platform without substanaive change. Obsolescence of software is more whim and fad than function.
On the point of ownership, software is considered a “writing” under copyright law. The original author gets copyright by convention. So, software, unless otherwise “licensed” gets the same protections as the Hobbit, Mickey Mouse, and Frank Sinatra’s recordings (all infinite if the Intellectual Property lobby gets its way). Thus the distinction between “Open Source” licensing that effectively puts the works into the public domain and the licensing that retains rights protective of proprietary (pecuniary) interests. Proprietary software (licensed under proprietary licensing terms) should be valued as any copyrighted works on some cost-basis.
Thanks for the feedback! Though I was careful to say “most commercial software,” any F/OSS software under GNU/GPL or similar non-proprietary licensing is still not capital.
Yep, agreed – there’s been a symbiotic relationship between software and hardware, until now. Just about the time hardware would reach end-of-life, there’d also be a reinforcing end-of-life refresh for software. Or vice versa. The dependencies were built-in as part of the business model for both soft- and hardwares.
Of course Microsoft would argue they exited the oligopolistic control of desktop environs, under governmental pressure to end anti-competitive practices.
Thanks for the feedback!
Thanks Rayne. I do not really know the technical definition of capital. But I do think controlling platforms is crucial to making money in computer technology. For a recent example, that is just what Facebook is trying to do by hosting newspapers such as the NYT. You may have noticed that CISCO is struggling a little. That is partly because of Snowden, but mainly because no one can really control the router platform as the TCP/IP standard is totally open. Licensing is important but not everything. ARM Holdings licenses out the ARM instruction set architecture, which is used by practically every mobile device, but their profits are puny.
Interesting, that about TCP/IP, had forgotten about openness due to its origins as a government-funded project. Cisco appears to have missed factoring risks inherent in using open intellectual property. On the other hand, who would ever have figured in the 1960s that DARPA’s ARPANET protocols would be used so widely, or blowback against the public when used by criminals and their own government?
I too skimmed over Rognlie’s critique of Piketty and his point about software struck me as seriously wrong. I would make two points about software.
The value in software for companies like Microsoft and Google does not actually lie in the lines of program code. These companies generate tens of billions of profits by controlling a platform.
Another related point is the many complex ways in which dependencies develop between users and computer software as well as between different pieces of program code. These dependencies are very hard to break or realign. It is because of these dependencies that Office (which has always struck me as a piece of trash) remains a cash cow for Microsoft.
ESRI is another big software company – they license the major GIS program, ArcGIS, which itself has add-ons licensed by other companies (we used DFDM from Geofields).
Yup. Licensed products, again, not owned. I wonder, too, how much of those GIS products are also impacted by public investment in OpenGIS, if they are compliant products? Can’t be entirely private capital if so.
They have this table listing standards and whatnot that they support. I’d expect them to be involved with OpenGIS, just because they’re the 800-pound gorilla in that field.
This is essentially an argument about ‘intellectual property’ as ‘property’ which is a load of crap. For a daily discussion of these issues see TechDirt (not my site so no conflict of interest).
Nice post Rayne, sorry to be late to the party, been busy. Not sure that it is appropriate for Rognlie to compare to different technology from a different era, with the different era being the more important. Nothing changed very quickly in the middle ages of manufacturing.
.
A more appropriate comparison might be to the hardware software runs on these days. You know, capital items with a 3 or 4 year functional life; it’s still warm when it hits the dumpster, made obsolete by the inexorable operation of Moore’s Law, year in year out, over the last 50 years. Computer depreciation schedules tend to be longer than functional life, so hardware often gets lump sum remaining undepreciated balance writeoffs at the end. As others have noted, software also commonly bridges generations of the hardware it runs on.
.
In addition to things like office suites, browsers and OSs, the apps my company makes filled a niche in the late 70’s when we developed them. They still do. We’ve outlasted close to a dozen hardware generations and several platforms. With browser based clients we’re pretty much hardware agnostic these days. Our software has the longevity of Rognlie’s apocryphal metal press and is a real live anecdote.
.
As you note, whether for profit or non, software is almost always licensed/leased, not sold.