But since the social marginal cost of botb DR-DOS and MS-DOS (and indeed of all OSs) are equal (and nearly zero), this change in the ratio of MS-DOS per PC does not introduce an inefficiency.
This appears to be true in the case of MS-DOS and DR-DOS since the production cost of both MS-DOS and DR-DOS are the same, while DR-DOS is of arguably higher quality than MS-DOS.
While this may be privately profitable, the social gain is zero, even if it did not induce the exit of rival operating systems such as DR-DOS with its attendant expected effects on raising the MS-DOS license fees.
Especially at the currently low royalty rates that would continue with competition from DR-DOS, few, if any, of the major OEMs would have any incentive to defraud MS.
This implies that, since the Microsoft royalty rate rose steadily until the availability of DR-DOS induced a sharp decline, the prevalence of CPU licensing should have increased over time until DR-DOS appeared, and then declined.
Finally since, as noted above, an OEM's incentive to underreport is proportional to the size of the royalty, and since the availability of DR-DOS has resulted in a sharp decline in Microsoft's royalty, it would appear that maintaining effective competition in the OS market has a desirable by-product--it reduces the extent of OEM fraud.