Initial and Annual Costs and Peacetime Life Expectancies
RAND's present convention in system costing adds the total of initial costs to four years of annual peacetime costs. The conventional character of this procedure has been frequently discussed. The equally arbitrary character of the total of initial cost itself, so far as I know, has not. Nothing, it appears, can be done very easily about the former arbitrariness. The latter, which has a sizable influence on certain comparisons, may be somewhat more corrigible.
Summing initial and annual costs: wartime difficulties.
a. Incremental first costs and wartime salvage.
Where a system element, such as a base installation, exists already, it need not be built new, but perhaps merely improved. Only these incremental costs, it is recognized, should be included as economic costs. Similarly, when the campaign objective of various systems being compared has been achieved, some of the surviving elements of these systems may be used in later campaigns. Its salvage value should then be subtracted from the system's costs to achieve the given objective.
The incremental initial cost can be estimated in a sensible way, though with a good deal of labour and some uncertainty, by analysis of existing facilities and their suitability. (I understand this was done for the Defense Study.) The problem in the case of wartime salvage value is to define the alternative uses and the probability of their occurrence. This might be done. Limited and rough judgments are very likely implicit in our weapons selections. But analysis of such judgments is very complex and the judgments are hard to confirm. It involves a degree of military clairvoyance so far not easily obtained.
b. Time of outbreak of war.
It is familiar (e.g. see missiles/Aircraft again) that the soundness of the practice of combining initial costs with four peacetime years rather than, say, two — as a prediction of expenditures — depends on when the war starts. If it starts at the end of the second peacetime year, four peacetime years will have been erroneously added. Actual systems costs will have been less. This affects the comparison of systems like missiles, which have low annual costs, with aircraft systems which have high ones. But here again clairvoyance, (in this case political as well as military) is so far lacking. (See, for example, on this, Project Delphi.) In the absence of clairvoyance, the missiles/aircraft study, as I understand it, is sensibly keeping separate tabs on initial and annual costs.2. The total of initial costs: peacetime life-expectancies.
Taking all of the elements of costs labeled "initial" (base, aircraft, personnel training, initial stock pile, etc.) summing them and combining the result with four years of costs which recur in every peacetime year amounts to the same thing as amortizing each of the initial costs in four years. Quite apart from such puzzlers as war and the time of outbreak, this practice is dubious. For these are subject to somewhat less precarious approximation than probabilities of war and campaign development.
Combat aircraft, it appears, do get replaced, if not exactly every four years, something like every five. Base installations, on the other hand, are used much longer. It's a fair bet that most of the bases now in operations will also be in operation 10 or 15 years from now if war breaks out at that time or if the preparedness program has continued up to then. This is not a question, then, of salvage value for wartime alternative use. These bases will almost certainly have peacetime "salvage value" — that is, continued operations in the same use — long after aircraft purchased at the same time are unfit for combat. Adding total base costs like total aircraft costs to the four year peace costs is quite misleading. The bias, moreover, is of some importance since base costs come to perhaps 20 or 30% of initial costs as presently calculated. They rank second only to aircraft. However harmless the bias might be elsewhere, this is obviously an essential matter to be considered in the study of bases, where a key question is precisely the rate at which one can trade a base expenditure for savings in total systems costs. This rate will be quite different, if the base expenditure is amortized in four years or in four times that number. It might also be of some significance in any systems analysis in which the systems differ in the proportion of installation to total cost.
3. The principle of separation into initial and annual charges is not entirely clear. The criterion, apparently implicit, concerns the period of recurrence of the expenditure under peacetime conditions. This, of course, is a matter of some uncertainty. In the case of base installation it is true that a drastic technological innovation might shorten the period of their use. It hardly seems likely, however, that complete substitution would occur in anything like four years. If missiles should be phased in as rapidly as their most optimistic analysts hope, it would still be over ten years, and even then, except for location changes in the case of the still more remote intercontinental missile, base facilities would be, to a considerable extent, convertible. The coming of the long-heralded truly intercontinental bomber, or the complete success of aerial refueling, if they should be cheaper to operate from home bases than systems using overseas bases, would outmode the overseas bomber base. However, if all of this does happen, it will change base requirements considerably more than four years hence.
In industry the significance of capital measurements is always a difficult matter; it swings between the trivial certainties of convention and the precariousness of technological and market prediction. In practice it has frequently formed the battle field for the Bureau of Internal Revenue and a business. Nonetheless rough judgments are made daily which assign considerably longer business life expectancy to real estate than to machinery and equipment, and to machinery and equipment than to office supplies. We would not avoid judgment by adding up the items of capital cost without discriminating differences in their life periods. This would amount merely to a belief in their equality which is, on the whole, less likely than the customary judgment. Similarly it appears that in our system analyses more reasonable estimates might be made of the appropriate peacetime amortization periods for various capital items than that of blanket equality, and in the comparisons themselves it should be fruitful to keep an eye out for sensitivities to whatever amortization assumptions we make.
All of this is under the assumption of peacetime conditions. I said earlier that these considerations are "quite apart from Puzzlers like war and the time of outbreak of war." In a sense, they are: in the sense that these are peacetime use-expectancies, easier to estimate than wartime ones. In another sense they are not because the duration of peacetime preparedness determines whether there will be time, in fact, to use these items. I think the very uncertainty as to the time of outbreak is what gives the peacetime assumption its importance. It makes it a dubious procedure to maximize preparedness for one point of time. The goal of the Air Force is more nearly to maximize its average preparedness over time. This supposedly is the best steady deterrent. Any such procedure must take into account the unequal life periods of various items which last beyond the first date of interest. As we have said, these life periods are uncertain, but by comparison with estimates of the time of outbreak of war, a great many things may be said with safety. One such thing is that under peacetime conditions bases will be used considerably longer than aircraft.
This note has confined comment to the installation and aircraft items of initial costs. This has been done both because the note was stimulated by problems raised in the base study and because the actual differences in life periods of initial items will probably weigh more heavily here than elsewhere; however, comments are in order also on other initial items.