N.I.S. or “not in stock” is the time honored phraseology of the Navy to indicate that an item requested is temporarily exhausted. It is the supply officer’s equivalent of the civilian grocer’s “fresh out,” the personnel officer’s “no one available,” or the housewife’s “I forgot to go by the drug store, so we’re out of toothpaste.”
Most officers, staff as well as line, feel frustrated and resentful when told their material is N.I.S. The well-nigh universal feeling is that the supply agency has fallen down on the job. Sometimes the lack of a critical item may spell the difference between success or failure of an important mission, and there is an understandable and very deep-seated concern that any supply activity that marks a requisition for soap dishes “N.I.S.” today may well mark another requisition for guns or ammunition “N.I.S.” at a critical time in the future.
The “N.I.S.” problem has been with us since the time of John Paul Jones, and it will most assuredly be with us long after the Navy moves into the promising future of guided missiles and interplanetary space. I propose in this brief paper to examine a few aspects of the problem and to suggest some of the things that will have to be done before it can be intelligently and rationally handled.
One thing to be understood at the outset is that it is useless to expect any supply agency to have 100% of stock on hand at all times. In fact, it can be demonstrated mathematically that if either receipts or issues (or both) are unpredictable, then it would require an infinite inventory to reduce the N.I.S. rate to zero. A slight amount of reflection will indicate that the problem is really one of degree: what are the limits of unpredictability? What slight N.I.S. rate can be tolerated? How much inventory can be economically handled?
The economics may be illustrated by the following anecdote. I am informed that one of the nation’s most distinguished economists, while working on a logistics research problem for the Bureau of Supplies and Accounts, visited a large New York department store to study their inventory control system. At the conclusion of his study he asked the top management of the store this question: “What is your N.I.S. rate in the white shirt department?” After consulting his control records, the store executive replied “about 2%.” Whereupon the economist informed him that the actual rate was 16%, not 2%. This news considerably agitated the executive and he stated in unmistakable terms that the department head would be called on for an explanation and would be directed to bring the rate down to 2% immediately.
The economist suggested a more cautious approach, and with the aid of data collected during his study was able to demonstrate that if the department head actually complied with the store policy, and reduced his N.I.S. rate to 2%, then the department would lose money. In other words, profit and goodwill on a few extra sales would not nearly compensate for the increased expense of purchasing and stocking the additional shirts.
This matter will be clearer, perhaps, if we compare the family refrigerator to a naval supply activity. The housewife normally makes, say, one or two trips to the grocery store a week and on each trip purchases a list of items which she anticipates will be needed by the family during the ensuing few days. So long as consumption by the family goes along uniformly and in accordance with her plans, nothing will become N.I.S. Suppose, however, that her spouse unexpectedly brings home several friends for dinner. All the butter may be consumed, and next morning the family’s toast is dry—butter is “N.I.S.,” and will continue to be until the housewife can make another trip to the grocer to replenish.
This example is tediously simple, but it serves to illustrate two additional and important points. First, no matter how full the refrigerator is stocked, there is always some conceivable emergency that will not be provided for. Even if plenty of allowance were made for unexpected guests, there would still not be enough on hand to carry the family through a major snowstorm or a strike of grocery clerks. The second point is that had the housewife been apprised of the husband’s plans, extra food would have been purchased and the butter would not have become an N.I.S. item.
Similarly with the Navy. We must exercise extremely keen judgment in deciding what contingencies to provide for. There are not enough warehouses or supplies in the world to insure against every eventuality. Furthermore, the logistics planner simply cannot do his job unless he is privy to war plans. We learned this lesson the hard way in World War II, as Ballantine* so ably relates. It is not enough for the logistics operator to be told what to procure. He must have sufficient knowledge to be able to gauge the effects of time delays, substitutions, and the economic and military effects of various N.I.S. levels and items. Not only will this help reduce the N.I.S. level during both a single operation and a whole war, but it will also ensure that the items which become N.I.S. are not vital to the mission. While much has been written about the interrelationship of logistics and strategy, there are still occasions when there is a tendency to provide the logistics agency with a mere list of items and expect them to be provided without exception. Since this is usually an impossible task, and since the logistics agency has no way of judging relative importance of items, there is an understandable tendency to “furnish what we can and forget the rest.” Some of “the rest” may be critical, and mutual recrimination between supply agency and operating agency will result. Application of the World War II lessons will easily avoid this sort of thing. The logistics agency should not be expected to run to the strategic planner every time a question comes up: the sheer volume of administrative decisions and plain guesses that have to be made in procuring any large quantity of materials makes such a procedure unworkable. In short, the only feasible solution is to ensure that the logistics agency has full access to strategic and tactical plans.
Turning now to some of the more detailed aspects of the N.I.S. problem as the Navy faces it today, it is quite clear that the better we can predict receipts and forecast issues, then the easier it becomes to keep enough stock on hand to reduce the N.I.S. rate.
Why is supply (or receipts) unpredictable? Many glib articles have been written about “lead time”—the time which necessarily must elapse between the moment an order is placed and the moment the item is actually received. Most of the authors have tacitly assumed that “lead time” is a fixed, or at least a known, period . . . say six months for clothing, or three weeks for gasoline. In fact it is no such thing. Lead time is actually rather unpredictable for a very large part of items in the Navy Supply System. Take melton cloth used in blue jumpers, for instance. The Clothing Supply Officer could count on a lead time of nine months for this material in 1943, two months in 1946 when surplus materials were on the market, twelve to eighteen months in 1947 when woolen mills were booked to capacity, and six months in 1950 before the Korean war. The explanation is immediately obvious. Some articles can be procured in a month if they are already fabricated and available; but if they have to be manufactured, the lead time will vary inversely with the number of competent manufacturers who will accept contracts, and with the ease with which they can obtain raw materials. All of these things change from year to year and even from week to week.
Even if a normal lead time could be established for every commodity, there are still the variations caused by shipping delays, strikes, failure to pass inspection standards, and the thousand and one “acts of God” that rise to plague the best intentioned supply officer. Who could foresee in June, 1950, that certain items of electronics supply with a normal lead time of three months would within a short while have an actual lead time of two years?
Here again it appears that it is impossible to provide for all contingencies. There will inevitably be times when supplies do not arrive as planned. The best that can be done is gauge the economics of the situation—the extra cost of providing against more and more remote contingencies as compared to the gain from not being N.I.S. should the contingencies occur.
Predicting issues, or “forecasting demand” as it is known to the civilian business man, is unfortunately not much easier than predicting supply right now. Much can be done to improve this situation, and this will be discussed now.
There are a great many stock control systems used in industry and the armed forces which attempt to forecast demand by projecting past issues. Thus it is assumed that if ten brooms were issued during the past year, the probability is that ten brooms will be issued next year. There are other stock control systems which slightly refine the procedure by “fitting” more complex mathematical curves. For example, suppose five brooms were issued in 1951, ten in 1952, and fifteen in 1953, then a possible projection would be a demand of twenty brooms in 1954. Another possible projection would be ten brooms in 1954 (based on the average issues of five, ten, and fifteen during the preceding three years). The defect in all these systems is the defect in the housewife’s refrigerator logistics, if she is unaware of the husband’s plans.
The Bureau of Supplies and Accounts has been acutely aware of the forecasting problem and, in view of the limited funds available for the task, has made extraordinary progress in solving it. The real answer, of course, is to find the basic factors which cause changes in demand. Once these are known, they can be measured, correlated with past issues, and projected into the future to give highly accurate predictions.
Consider a piece of electronics equipment, for example. The Electronics Supply Officer is responsible for making available tubes for the equipment whenever they burn out. It is reasonable to assume that the life of tubes depends mainly on how often equipment is turned on and off and how many hours it is run. By analyzing planned operating schedules for the equipment, it becomes rather easy to predict how many tubes will be burned out—and the prediction will have a high degree of accuracy. This particular example of using planned operations to predict future requirements of spare parts is actually being studied now. As we learn more about the reliability of predictions, it should become possible to obtain overall factors for various groups of items, or possibly even for a whole supply system (e.g., the Fuel Supply System). But this is no place for snap judgment or specious reasoning; for the success or failure of many missions will depend on the accuracy of these demand predictions. Nothing short of the best answer that trained mathematicians and experienced technical and supply officers can produce will serve. The careful studies which are being made should be continued and encouraged; they should not be pressured into hasty decisions which could easily lead the nation to disaster.
I have outlined one method of forecasting which is being studied by the Electronics Supply Office. There are other methods, some depending on complex statistical sampling techniques, others on harmonic analysis of times series, others on more extensive and frequent demand reports, and so on almost without limit. The details need not concern us here. It is important to remember that all methods of forecasting demand are utterly dependent upon having complete information of the Navy’s operating plans. It is trite to say that one has to know how far and how fast ships will steam in order to have enough oil on hand for them; and yet, on a lower level, all of us known of situations in the past war where supply officers, not privy to some commander’s operations order, failed to provide the right type of clothing, or enough general stores hardward to repair battered small boats.
There is another aspect of this scientific forecasting well worthy of attention. Any supply system, civilian or military, is like a freight train: a push on the end car is not communicated to the engine until all the intervening cars are bumped in turn. The speed with which a push or pull on the end car is communicated to the engine is termed the “responsiveness” or “sensitivity” of the system. All writers on the subject are agreed that one of the primary goals of the Navy Supply System must be a high sensitivity. Supply Demand Control Points, represented by the engine pulling and guiding the string of cars, must be able to detect very quickly any change in the position of the last car, which represents the last issuing supply activity.
How rapidly can these changes be detected now? The answer is not encouraging. Under the best of conditions, using modern electric accounting machines and punched card reporting, it takes about 45 days from the time an issuing activity can determine its inventory until the Supply Demand Control Point can process the punched cards, analyze all inventories in the supply system, and decide how best to redistribute stock and how much new stock to procure. This 45 days, of course, must be added to lead time, and in actual dollars and cents represents an enormous investment of goods “in the pipeline.” Furthermore, it means that a minimum of 45 days will elapse before the Control Point becomes aware of a sudden increase or drop in demand.
The Bureau of Supplies and Accounts has long recognized that from a strictly business viewpoint it would mean great financial savings to reduce this time. These savings would come from needing less goods in the pipeline, from having less physical material to handle, and also from reducing the N.I.S. rate and hence the volume of paperwork and processing. Hence there has been a two pronged attack on the problem.
The first approach is to assume we must retain the present method of reporting and forecasting demand. How can the processing be speeded up? One way is by using high speed electronic computers with extensive memory circuits. Such devices will reduce the time for processing demand data and should pay for themselves many times over. Meanwhile computers which can be better fitted to the logistics problem than presently available instruments are being actively investigated under the supervision of the Office of Naval Research. Within a few years it should be possible to obtain instruments designed especially for forecasting and inventory control work, and once installed at the various Supply Control Points we shall have gone a very long way indeed toward reducing the pipeline requirements.
The second approach also assumes we must retain the present method of forecasting, but speeds up the processing by changing the method of reporting. As experience has been gained with the integrated supply system, it has been found that some items are more unpredictable than others. We are more likely to have a sudden “run” on jet aircraft engines than on pencils. Hence it is possible to classify items on the supply system and require different methods of reporting for them. Critical items like aircraft engines might be reported once a week or even once a day, other items like pencils (having a sustained demand and of a non-critical nature) reported only annually. General Motors installed a system somewhat similar to this in its spare parts distribution system and has obtained excellent results in increasing the sensitivity of the system and in reducing the N.I.S. rate. The Bureau of Supplies and Accounts is very actively studying this type of approach to the overall problem and may well adopt some of the techniques even if totally new methods of forecasting are developed.
In summary, the N.I.S. problem will always be a “problem” since there will always be a few items not in stock in any normal supply activity handling thousands and thousands of items. We can reduce the number of N.I.S. items, however, by furnishing logistics planners with all available information on future naval operations. We can also reduce the N.I.S. items by improving the sensitivity of the supply system. The most promising method of doing this is by changing our forecasting methods—by predicting future issues on the basis of planned employment of fleet units rather than on the present basis of past issues. However, even if improved forecasting methods are not used, the N.I.S. rate can be reduced by speeding up our present data processing with high speed electronic computers or by improved reporting systems.
Now for the crux of the problem. Let us assume that the intelligent and aggressive studies being conducted by the Bureau of Supplies and Accounts are completely successful. The Navy will then have available (for the first time in the history of any supply system in the world) a set of data that will show us what it costs to achieve any desired N.I.S. rate. The data may be conveniently presented by comparing the two curves at the top of this page.
At this point a command decision must be made: “what N.I.S. rate can we afford.” A low N.I.S. rate will require diversion of war materials and manpower which, perhaps, are more urgently needed elsewhere. A high N.I.S. rate will be cheap in money, but will obviously reach a point at which it endangers the success of tactical missions. There are advanced mathematical techniques (games theory) which can further narrow the area of command decision, but these considerations are beyond the scope of this very brief resume of the problem.
Getting to the point where we are sure of an optimum grand strategy—of the best military employment of our supply system— is a long way off. The biggest problem, as usual, is dollars. Research funds for the type of studies needed are very hard indeed to come by. The nation is all too willing to put up millions or billions of dollars for development of a new weapon. But it is very difficult to convince able and conscientious men sitting on the “wrong side of the table” that a few hundred thousand dollars spent on a theoretical mathematical study is worthwhile. Supply Corps officers know intuitively that a good supply system will double or quadruple the actual firepower we can direct against an enemy—and this is usually far more benefit than can be expected from any new piece of “hardware.” It seems clear that we must resolutely direct our research dollars into a more proper balance between “hardware” and “logistics” projects than has existed heretofore. There are many aspects that have not been covered.
In conclusion, I wish to reiterate that the foregoing presentation has been greatly simplified to achieve clarity. Actually the problem is not an “N.I.S. problem” but one of maximizing critical support functions and minimizing costs (economic, vulnerability, and others). The problem is complex, one requiring extensive mathematical treatment as well as keen professional insight, and one which must be recognized, and recognized with maturity, by the Navy.
* Ballantine, Duncan, U. S. Naval Logistics in World War II. Princeton, N. J.: Princeton University Press, 1947.