[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re #3: Scale Economies at Microsoft?

  I'm afraid I was a little loose with the math when I responded to this
  before (at 2 am or so), so let me put it a little more formally.  Call
  the total cost of development of a software product dc, the number of
  units sold nu, and the selling price per unit sp. The cost per unit is:
  There, the profit per unit sold is:
  sp - (dc/nu)
  Total profit is profit per unit times the number of units, i.e.,
  nu(sp - (dc/nu)) = nu*sp - dc
  Notice that as sales go up, nu increases but dc does not.
  The concept of predatory pricing doesn't really apply since the concept
  of per unit cost doesn't, at least not as traditionally concieved. The
  per unit cost depends on the number of units sold. Each new copy of
  Windows 95 sold increases the amortization of the development costs and
  thereby reduces the "per unit cost", not only of itself, but of all the
  previously sold copies. 
  I've included responses to particular points below.
  charles mueller wrote:
  >         I have a private communication advising me that the software
  > industry, in its entirety, is inherently monopolistic because of scale
  > economies.  There are 2 costs, I'm told, (1) development, and (2)
  > reproduction and distribution.  Only the first is expensive.  The second,
  > reproduction and distribution, is virtually costless.  The industry is
  > divided into a series of "niche" markets, all of which "rapidly evolve
  > toward monopoly," since the leaders in all of the niches "can defend their
  > market share by reducing their prices to practically nothing, making price
  > competition SUICIDAL for newcomers."  The goal of all firms in the software
  > industry is the same--"to gain a dominant position in a NEW niche market.
  > There is essentially no new investment in EXISTING niche markets, since it
  > is impossible to compete with an established dominant player on the basis of
  > lower costs...."
  >         Accordingly, antitrust is powerless to make the software industry
  > competitive. Monopoly is its nature.  One niche, one monopoly.  OS, Bill Gates.
  >         Scale economies--the savings (in unit costs) that come with higher
  > volumes of output--are typically exhausted rather rapidly in most
  > industries.  The common estimate in the auto industry, for example, is that
  > there are no cost savings beyond 7.5% of the U.S. car market.  At that
  > volume, unit costs are as low as they can be driven with the existing
  > technology.  There are virtually no industries in which a market share
  > greater than 10% of the U.S. market is required to exhaust all scale
  > economies--to reach the lowest (technologically) attainable cost per unit.
  >         In addition, those unit cost savings, even when real, are often
  > trivial beyond some modest output level.  If a widget costs $1.00 when made
  > by a firm with sales of 1 million units per year--versus $1.01 when produced
  > in a volume of only 1,000--the competitive significance of that 1%
  > difference is likely to be close to zilch.  In U.S. industries, the
  > lowest-cost firm--the one the others freely acknowledge as their industry's
  > most efficient--is typically one of its smallest (and newest), with a market
  > share of 5% or less.  (See my Web site, below.)
  >         Is the computer software industry an exception?  My understanding is
  > that the costly part (development) culminates in a "master" disk of some
  > sort.  For example, computer manufacturers such as Compaq and IBM, after
  > paying Bill Gates' high "licensing" fee, receive such a "master," from which
  > they can, in turn, produce the number of copies needed for the millions of
  > machines they will be shipping to consumers.  So there are no significant
  > software scale economies at the computer manufacturing level.  The latest
  > reported figures I've seen, for example, put Compaq at the top with a market
  > share of some 12%, hardly a figure reflecting heavy scale economies.
  Compaq is a hardware, not a software company. Computer manufacturing is
  closer to traditional industrial economics as there is an inevitable
  per-unit cost. The more R&D heavy areas of hardware are somewhat in the
  middle: yes, Intel does have an inevitable per-unit manufacturing cost,
  but their R&D cost exceeds this, with visible results. 
  >         Where, then, are the big efficiencies in the software industry that
  > condemn it to only 1 firm in each of its various "niches," e.g.,  Bill (and
  > Bill alone) in OS?  How big are the savings, if any, to the consuming public
  > from Bill's monopoly?  How large are the extra per-unit costs, if any, where
  > a 2nd, 3rd, and 9th firm appear?   
  A new firm has to cover its entire R&D cost with a smaller number of
  units sold (initially, this is bound to be the case).
  What are Bill's total costs for a copy of
  > the "master" he sells to, say, Compaq, et al?  How many of those "masters"
  > does he sell?  How much does he charge for each?  What do they cost him?
  The master disk as a physical entity is irrelevant. The value is in the
  reproducable information contained on it. 
  > How much does his per-unit cost differ as between, say, 5,000,  50,000, and
  > 100,00 units?
  See above.
  >         My correspondent describes a software world in which all "price
  > competition is suicidal for newcomers."  With essentially zero unit costs,
  > the dominant firm in each "niche" market can easily wipe out any newcomer
  > "by reducing their prices to practically nothing."
  >         Once upon a time, that was called predatory pricing--selling below
  > cost to destroy a competitor.
  Not really relevant, since the per-unit cost is determined by the number
  of units sold once the development costs are taken as given.
  >         Can the group help us out here?
  >         Charles Mueller, Editor
  >         http://webpages.metrolink.net/~cmueller
  >         Charles Mueller, Editor
  >         http://webpages.metrolink.net/~cmueller