Re: Merge Intel & Microsoft

Ross A. Finlayson (
Fri, 01 Oct 1999 19:11:43 -0400

Robin Hanson wrote:

> Robert Bradbury wrote:
> > > Intel and Microsoft both seem to have strong near-
> > > monopoly market power, and their products are clearly
> > > strong complements. So the standard industrial
> > > organization analysis would suggest that merging them
> > > would produce lower prices *and* more profits for the
> > > merged firm.
> >
> >... you may be speaking at an economics level most
> >people don't have. ... only way I can interpret this statement
> >is that if you mean that merging the two corporations makes
> >a more efficient corporation by reducing management overhead
> >... In short, I think you need to explain the case better so
> >people can see what you see.
> I wasn't trying to defend or explain the standard claim, just
> presuming that as in all conversations those who understand
> the topic might join in, and those who don't would stay out.
> We have conversations all the time here about topics many
> subscribers must be clueless about.
> But in this case, I'll explain. Consider a product made of
> two parts, X,Y, such that there is no point in having one of
> the parts without the other (e.g. computer chip and OS). The
> quantity demanded D(P) of this product is therefore in terms
> of the price P of the two parts together. If the two parts
> are sold by separate monopolists, then P = PX + PY and each
> monopolist does something like
> max (PX - CX) * D(PX + PY)
> PX
> Where CX is the cost of making X. If the two firms are
> merged, then together they
> max (P - CX - CY) * D(P)
> P
> This leads to both a lower price P, and more profits. QED.
> Robin Hanson
> Asst. Prof. Economics, George Mason University
> MSN 1D3, Carow Hall, Fairfax VA 22030
> 703-993-2326 FAX: 703-993-2323

Well, no matter what price for chip and OS, here X and Y, saying that the rest of the case and cables and storage are marginal, then still a P(X+Y), that is price of chip with bundled OS, that yields the highest quantity D(P) at price exists.

These cases etcetera are being manufactured, the chips added, and bundled OS installed by a large variety of OEMs.

If the two firms were merged, and selling their combined product (one OS license per chip) then the price of the combined product would still have the same demand for a given price. Thus, the price components as they contribute to gross income would still sum to the same as if there were separate prices for chip and OS. The price as gross income could be redistributed among the operating components of the two merged industries. That is, D(PX+PY)=D(P), and P=PX+PY and C(total cost)= CX+CY. Thus the sum gross income per unit equals PX+PY-(CX+CY), which equals P-(CX+CY), for a given price P that yields D(P) to maximize P*D(P).

It is true that a price could be in agreement between manufacturer of X and Y that would yield max P. Otherwise we assume Smith's "Invisible Hand" does this. Of course, that carries its own assumptions like perfect information.

I think you are stating that if the two companies were run as one then they would offer a lower price P(X+Y) to get a higher demand than individually, ie P(X+Y)<P(X)+P(Y), D(P(X+Y))*P(X+Y) >= D(P(X))*P(X)+ D(P(Y)) *P(Y).

As it is we are seeing a bunch of OEMs doing that, as they bundle the OS, chip, and other computer components, and computer prices drop all the time. These OEMs compete, as Intel and Microsoft get fat.

Also, there are issues of price elasticities that are different for X, Y, and X+Y.

Of course, demand depends on other factors besides price, for example perceived need and advertising or peer driven demand. Also, it is very likely that if one operating system is purchased bundled with a computer system, then any further software acquired would be to run on that operating system.

An interesting thing about software "manufacturing" is that CY decreases for each unit, that is, there are economies of scale. Once one copy of the operating system or other program is created, then any number of duplicates are marginal in cost. This is not so with chips, where each chip has certain resources required. Of course, there are still economies of scale in making chips. So, the only goal with software is to increase the number of copies sold, as there is almost no cost, in this case C(Y), thus C(Y) decreases for each copy sold, while C(X), after the fabrication plant is paid for, has more constant cost. So, if Microsoft and Intel were combined, then the price point that guaranteed the largest price times demand for chip, with price of OS to be approaching zero, would sell the most chips, and thus marginalize the cost of the fab, which is assumed to be much larger than the cost of an operating system. When the cost of the operating system is comparable to the cost of a fabrication plant, then positive costs of each would be weighed into the calculation of bundled price.

If Microsoft and Intel were to merge, there would be the possibility of gaining value from eliminating duplicate products, for example, Microsoft makes a compiler and Intel makes a compiler (that runs on Win32 OS). Generally, I know of few or no competitive Intel vs. Microsoft products.

Five or more years ago Windows/DOS was about the only OS for Intel architecture. Today, there is, for example, Linux, FreeBSD, BeOS, Solaris86, and others. In large part, source code for any one of thesse works transparently on any of the others. Of course, the standard C library exists on Windows as well, but in terms of windowing architecture when speaking more of X Windows vs. Windows, X Windows apps are portable to more operating systems.

So, to the question of would merging Intel and Microsoft be good for Intel and Microsoft, in large part it would because they could be even more insular against the other chip and OS manufacturers. At issue is whether combining would be good or bad for operations, either good by combining efforts and direction, or bad by promoting bloat and sloth, or more realistically how much of either would take effect.

I do admit, I use daily a computer running Windows on an Intel chipset.

It's a question that is valid in the sphere of large scale daily use of computers by many, much more so than, for example, if apples and oranges should merge. Certainly, the large scale pricing and distribution models of chips and software is one of which I know not much, but these are my words about it.

Have a nice day,

Ross F.