No, I didn't just whip this up, it's something I first wrote in 1997 and pull out for appropriate occasions. I've noticed that the SPA/BSA/ICCS reports are now much thinner on methodology, though their grossly inflated loss claims persist.
After publishing this, I checked with Hal Varian (refernce below), now Dean of the School of Information Management Systems at UC Berkeley and a highly respected economist to see if I wasn't completely out to lunch. "Looks about right" was his response. The Register has also run a number of pieces on software piracy and enforcement actions, most of which are strongly critical of the BSA and Microsoft's tactics in this arena. I'll try to post some recent links in this discussion. Many of the predictions in this piece have been borne out.
Much of this material was originally posted by me to
misc.legal.computing following Rachel Guerin's posting of a summary of the
SPA's release of 1997 Global Software Piracy
In this post, I challenge both the methodology used in estimating business
losses due to piracy, particularly the dollar revenue losses, and
several implications of piracy, based on economic analysis.
Rachel Guerin wrote:
Questioning the study and the numbers may be intellectually
stimulating, but the fact remains, theft is theft. As a society we
have deemed stealing wrong. As such, we need to enforce that standard
for all property, intellectual and otherwise, a responsibility which
falls to governments.
We should also be concerned that theft is reported accurately. I
wouldn't want to see $0.50 of plastic beads claimed as a $10,000 pearl
Finally, stealing a car from the dealer, not owner, is more analagous to s/w piracy. As such, thay have lost a sale and someone has gained from that loss. Just because copying s/w is easy, doesn't make it right. How many people would walk into CompUSA, pick up a s/w package and walk out the door without paying? Now, how many people would copy their friend's titles? It's the same thing.
It's not the same thing. In the first case (walking out the door without paying), I've removed property for which the store has paid, without paying in turn. In the second, I've entered into competition with the store, with a much less expensive product. Illegally, we'll both agree. But not the same thing.
Economic Supply and Demand
As has been pointed out with distressing regularity when discussing software piracy, pirated software does not represent a redistribution of already manufactured goods (I'm not stealing your car), it represents production of additional, unauthorized goods (I copied it). What I haven't seen stated often is the correct economic description of what piracy is and how it operates.
The following discussion assumes a basic understanding of microeconomics.
Economic "demand" is the units of a product which will be purchased at a given price. Plotting price on the vertical and unit on the horizontal axis, this is usually a downward sloping curve, and tends to become flatter to the right. Economic "supply" is the units of a product which will be produced at a given price. This is an upward sloping curve, and tends to steepen to the right. These curves are a simple, but effective way of describing complex underlying production and consumption functions and preferences. See any econ or microecon text, e.g.:
Samuelson, "Economics", Hal Varian, "Intermediate Microeconomics".
Software piracy is economically equivalent to introducing an almost identical, substitutable product on the market. By substitutable, I mean that a single consumer is unlikely to buy both the pirate and legitimate version of a program. By almost identical, I mean that pirate software is (usually) functionally identical to the legitimate product, but may have some differences in perceived benefits (manuals, support), and risks (criminal liability, viruses).
The two products have independent supply and demand curves (S-D curves). The relationship of these curves determines the real effects of piracy.
Pirate supply is right-shifted from legitimate supply (more units produced at a given price). Pirate demand is usually left-shifted from legitimate demand (fewer units demanded at a given price. Why the last? Think: at a given price, you can purchase either a legitimate, supported, and (presumably) virus-free copy of software. How much less would you be willing to pay to lose these assurances?
The result is that pirated software is always less expensive than legitimate software. How many units of each are sold depends on the relative shapes of the S-D curves -- pirate sales may be higher or lower than legitimate sales.
The effect of pirated software is to reduce the demand for legitimate software. The result of reduced demand is to move back along the legitimate software supply curve. The result is nonintuitive: software
piracy reduces the price of legitimate software. Again, the amount varies with the particular S-D curves. (We'll get to revenue impacts later).
The empirical evidence is strong. In markets with high piracy rates, legitimate software is significantly discounted. This is the case in Hong Kong and other eastern markets, where legal copies of MS Office
sell at 50% or greater discounts to prices in the US or western Europe. Pirated software sells for pennies over the media cost. A colleague was telling me of his experience at the Golden Market (?) in Hong Kong --
separate pricing applies if you supply your own media. Piracy is a highly competitive market, driving profit margins down.
Incidentally, if you read carefully, the SPA's report never touches this topic at all. Some derived news reports talk about "increased costs to consumers", but these are largely fluff pieces written by reporters
with, presumably, less knowledge of economics than the SPA.
Impacts on ISVs
Piracy does have a negative impact on legitimate software revenues. The amount cited by the SPA as "losses due to piracy" is mislabeled, it is really the wholesale value of pirated software (units of pirated software times wholesale price).
The actual loss is the difference between what revenues would have been without piracy and actual revenues with piracy. Determining the price
and unit sales in the absence of piracy requires a more rigorous economic analysis than is used in the SPA case. The amount is far, far less than the numbers reported by SPA. I've been looking for, but have been unable to find any rigorous economic analysis suggesting the what
the actual amounts might be in different markets.
It's possible to get some idea of how far off these are simply by backing out SPA's numbers, taking the piracy rate to represent a fraction of total wholesale unit sales. Taking a few of the markets described, 1998 SPA report:
(billions of US dollars -- and curse Scoop's lack of table support....)
Market : Whsl Value Pirate Sw : Piracy Rate : Legitimate Sales (my estimate)
- US : $2.8 : 27% : $7.6
- China : $1.4 : 96% : $0.06
- Japan : $0.8 : 32% : $1.7
- Korea : $0.6 : 67% : $0.3
- Brazil : $0.4 : 62% : $0.3
- UK : $0.3 : 31% : $0.7
China is of particular interest. Though piracy "losses" of $1.4 billion are claimed, ACTUAL software sales were only $60 million. I find it
impossible to believe that any significant fraction of the claimed billion plus dollars in "losses" would have been realized in actual sales, in the absence of piracy.
Put another way, the US per capita legitimate software purchase for 1997 was $27. For China, it was $0.05.
Critique of SPA Analysis
SPA is very misleading in its labeling of the components of its
- "Demand" by SPA is the number of computers (HW units) present in a
country or region based on sales reports, multiplied by a derived
software utilization factor whose derivation is unspecified.
- "Supply" by SPA is the number of actual software unit sales of
software, from vendor reports.
In economic terminology, SPA's "demand" is a measure of maximum
potential capacity or consumption. SPA's "Supply" is just the number of
SW units sold at prevailing prices. They have very little to do with
the actual economic concepts of supply and demand, which describe a
relationship between price and units.
As SW piracy is largely an economic crime (the contribution of "spree"
or "joyriding" or "a copy for a friend" SW piracy is negligible, though
exceeding licensed use within a firm is not), the best defenses are
likewise economically based. The SPA report and its derivative press
coverage are an attempt to influence the demand side of the equation --
raise doubts, concerns, and scruples to the point that demand for pirate
software is reduced.
The alternative attack would be on the supply component. If pirate markets are largely concentrated among a few marketplaces such as Hong Kong, it might be possible to undercut the market by entering the pirate
market and selling software below media costs. This "pirate" pirate shop would draw customers from "legitimate" pirate shops, drying up their market by pushing prices to unprofitable levels. This wouldn't
take much of a price cut, as the market is already highly competitive.
The effectiveness of this strategy would be strongly affected by such factors as the presence of the mob (requires additional security, may mean pirates have financial backing to wait out dry periods), the need
for continued ongoing sales, and the efficiency of the markets. If a high proportion of sales occur in a small number of markets, with poorly backed, unorganized pirates, and market reemergence takes time, this
could be highly effective. The less applicable these assumptions, the less effective this strategy.
Finally, there's the question of whether or not piracy is good or bad,
for both the software companies and society.
In the case of China, there is more than 20 times the availability of software with piracy then without. What we need is a cost-benefit analysis of this situation. Curiously, the benefits are easier to quantify than the losses. Presumably, each pirate copy represents some
productivity gain for the person using it. I don't know how to quantify the market efficiency loss, and I don't know that it goes beyond the real (not SPA stated) lost revenues of the software company.
Microsoft, in particular, has always had a relatively cavalier attitude toward capturing all potential revenues. Their business model has been based on the size of the market, rather than on the portion or efficiency with which revenues are collected, and has indicated a preference to capturing a small percentage of a large market. The telling comparision is with Apple, which long had a tremendous profit margin on its systems. With market share at 3% and falling, Apple's strategy loses.
I've seen discussions suggesting that in a poor but growing economy, even pirate sales are a benefit to the vendor. Software, like heroin, is addictive. Once a user is hooked on a specific brand, costs of switching, even if non-economic, are high. Legitimate software, while an income stream, is also a liability for support, warrantee returns, and related servicing costs. Pirated software is unsupported, and does not bear this liability.
Even in western countries, many people adopt a "try before buy" attitude to software, and may sample a product illegally before actually purchasing it. Additionally, I strongly question whether any significant portion of US pirated software would have been sold at current market conditions.
A Rock and a....
Finally, with the growing popularity and proven performance of free and open software alternatives, commercial software vendors may not have a competitive choice but to allow rampant pirating of their products,
simply to gain a market position, however non-profitable it may be. In Redmond's eyes, a million pirated installs of Windows NT may still be preferable to a million installs of Linux.
I'd say they're stuck between a rock and hard penguin.
Karsten M. Self
SCO -- backgrounder on Caldera/SCO vs IBM
Support the EFF!!
There is no K5 cabal.
[ Parent ]