L.S.E. Essays on Cost

4
The rationale of cost accounting
by R. S. EDWARDS

First published in Some Modern Business Problems, ed. A. Plant (London 1937).

I. NATURE AND OBJECTS OF COST ACCOUNTING

There is always a risk that a paper which deals with the general principles of a subject may be dubbed theoretical and on that account of little value to the businessman. It may be felt at the end of this lecture that I ought to be criticized on this score, and in anticipation I want to refer to the work of two of the earliest English writers on costing.

In 1887 Emile Garcke, an engineer, and F. M. Fells, an accountant, published jointly a work entitled Factory Accounts, and it would be generally admitted that the principles of costing which they set down remain largely unaltered to this day. Yet the review of the first edition in the Accountant was to the effect that the book was more theoretical than practical, that it was pedantic and involved 'in the nature of a work on political economy'. Moreover the claim that stock balances should be as readily ascertainable as cash balances was regarded as an unattainable ideal. Today the suggestions put forward in this book are regarded as commonplaces. I hope therefore that if anything I may say appears to be more academic than businesslike, it will not be condemned on those grounds alone, particularly in view of the fact that there is no time to develop practical applications.

Some familiarity with the general nature of cost accounting, as it is at present practised, will be taken for granted. No clearcut definition is possible, and one can always point to borderline cases and ask 'Is this financial accounting or costing?' or again 'Is this statistics or accounting?' Little is gained by such discussions; in general we have come to look on costing as an elaboration of accounting used particularly in those cases where a number of factors of production are combined in manufacture.*51 The financial accounts disclose the total profit of a period, while costing systems normally aim at explaining this figure. An endeavour is made to analyse the past in order to avoid mistakes in the future. I think, however, that we must be quite clear that costing does not and is not intended to provide data for major changes in policy. For example, decisions whether to open a new department or close an old one would require special cost investigations which the ordinary cost-accounting system could not cover. It is the constantly recurring management problems on which costing is supposed to shed light.

The principle question we have to ask is this: out of the mass of interesting data which is available in any business, how much is worth collecting? The answer would appear to be: only that which can influence policy. Unless the information supplied enables the management to do something or refrain from doing something, its collection is not a business proposition.*52 Costing fails in its object unless it adds more to net profit than the expense of running the system.

In this connection experience in France is interesting. Throughout the nineteenth century French writers were producing interesting books on cost accounting for agriculture—they were far ahead of us in this direction. In the preface of most of these works there appeared a lamentable tale about the farmer's tardiness in improving his accounting methods. In spite of these admirable works the French farmer continued in his old ways. The probable reason is that, interesting as the information might have been, it would not have influenced policy sufficiently to add enough to profit to cover the expense of keeping the accounts.

It is not always easy to find out whether the collection of certain figures will pay. There is a wide range of cost data which might lead to small changes in policy, but the influence of which is of doubtful importance. There is another range of costing information in the nature of laboratory work which must be regarded as a long-run investment. In general, then, our method of attack will have to be: firstly to decide what information we should regard as necessary, if its collection were costless, and secondly to consider whether this information is of sufficient importance to warrant the expense of its collection.

What policy-influencing information would one like to have if it were supplied free? In answer to this question one can introduce a useful division into what I shall call, for convenience only, the entrepreneurial and the technical problems. The businessman's entrepreneurial problem is to decide on the size or rate of output which will yield maximum profit. The technical problem is to produce this flow of output at the lowest cost possible for the given scale of production, having regard to the facilities available. In other words, the technical problem belongs to the works manager—he is told what output to produce and must produce it at the lowest possible cost. The entrepreneurial problem includes the technical ones, and is the managing director's province.

II. SOME DIFFICULTIES AND SUGGESTIONS REGARDING TERMINOLOGY

At this point we must stop to grapple with a few of the difficulties of accounting terminology. The main themes running through most books on costing concern firstly the difference between so-called 'direct' and 'indirect' cost and secondly the methods to be adopted in imputing the latter. That is to say, if aero engines are being produced, certain costs are traced to each engine or batch of engines and these are 'direct' costs. Raw material is the best example of this. But the great mass of overhead associated with the fixed equipment is regarded as 'indirect' cost and all sorts of ingenious attempts are made to allocate it. Broadly direct cost and prime cost are synonymous in accounting literature, and accountants would agree, I think, that either term includes those items which it is worth while to trace to the unit which is being costed. For analysis this is unsatisfactory, because it brings us to a position in which prime cost as a concept depends on book-keeping difficulties and one can never be sure what any individual writer means. Prime cost normally includes the relatively large items of traceable cost, but in practice it may contain more besides, particularly, as we shall see later, if a loading rate is added for machine time or to cover stores costs. The line drawn by accountants between prime or direct cost and overhead or indirect cost is therefore arbitrary. If half-a-dozen accountants were asked to split up the debit side of a manufacturing and profit-and-loss account between prime cost, works oncost, office oncost and distribution oncost, and were then asked how the last three were to be allocated to the products, there would be very wide divergences of opinion. This is bound to be so because more explicit assumptions are necessary. Firstly, what period of time is to be considered? Many costs which are fixed for a day may become variable in a week—a great deal of labour supply is like this. Secondly, what unit is to be costed? Is it to be a department or just one unit of product in that department?

The following is a typical statement on the subject under consideration: 'The cost of labour (i.e. the amount of productive wages paid) plus the cost of raw materials, with all charges thereon, such as carriage inwards, freight, dock dues, etc., is called the Prime Cost.'*53 Such a categorical statement is quite useless for our purpose. What are productive wages—do they include the works manager's salary? Where are we to draw the line between wages and salaries generally or between productive and non-productive labour? Are we never to include expenditure other than so-called productive wages and raw materials? The definition leaves the major questions unanswered; instead of describing the characteristics of prime cost, we are told that certain items are prime cost.

Now the most important thing about costs is the extent to which they change with output. Whatever unit one decides to cost, whether it be a job or a department, it can be said that if the job were refused or the department closed, the total expenses of the business would be less by an amount which could be determined fairly accurately.

That is to say, in undertaking the job or continuing the department, the business is involved in additional expenses which could be avoided. As a result, to be worth while, the unit costed must generally add to the turnover of the firm, at least the amount of this additional cost. To make the position clearer, let us suppose we are costing the production of a machine in an engineering works. If the order for this machine were refused, we should not have to buy or take from stock the materials for the job; it might be possible to dispense with some labour, and some wear and tear of machine tools would be avoided; perhaps the bank overdraft could be reduced and interest saved. On the other hand, the rent of the factory would remain the same, the foremen would still have to be paid, the machinery would become obsolescent at the same rate.

Many other expenses are similarly unavoidable. If we call the additional expense incurred by producing the unit to be costed 'variable cost' and other expense 'fixed cost', then we shall have a useful distinction, as the production is worth while if it covers the variable cost and contributes something, however little, to the fixed cost. There is nothing absolute about the term 'variable cost'; the items composing it will differ with the change in the unit.

It will be conceded, I think, that given the unit there will be little difficulty in collecting fairly accurately prime costs interpreted as variable costs. No theoretical difficulties will arise, the problems are those of book-keeping only. But there are many variable expenses the separate allocation of which is regarded as too expensive if a continuous double-entry costing system is in operation. These items are lumped into the overhead so that this includes some directly variable, some partly variable and some absolutely fixed costs. The enormous drawback to a double-entry costing system 'tied in' to the financial accounts is that some arbitrary assumptions have to be set up to make it workable. As Professor Canning has said, 'Cost accounting procedure rests upon oversimplified and largely arbitrary fundamental analysis'.

III. THE PROBLEM OF 'OVERHEAD'

All cost accountants will agree that their great difficulty is in the allocation of overhead or oncost. Naturally this is so, for a number of costs are being added together which have different degrees of variability.

First there are disputes as to what is manufacturing and what is distribution oncost, as it is assumed that these should be allocated on different bases; and whether certain expenses shall be left out altogether as being outside the scope of costing—cash discount and interest appear in this category. Then the expenses have to be split between departments; some are variable to the department and there is no difficulty, others are allocated arbitrarily on some basis considered 'fair'. Then comes the most difficult business of all—finding a way of 'tacking on' these departmental overhead costs to units of product which have gone through the departments. For example, overhead is allocated to the product variously as a percentage on so-called prime cost or on direct labour cost incurred or on the basis of direct labour hours or machine hours. Accountants condemn the 'prime cost' basis in most cases, and many of them object to 'direct labour cost', but on the whole they seem satisfied with the 'labour-hour' and 'machine-hour' rates. These are based on the generally implied assumption that overhead expenses vary with time. In the very long run this is probably true merely because in the long run most costs are prime costs; and practically everything wears out sooner or later. But in the relatively short period required for completion of a job or order it is not true.

To arrive at a percentage or rate for oncost distribution it is necessary to assume some output as a denominator. Given this estimated output and the estimated overhead for the period covered by the output, a rate can be established. We are usually told to use 'normal output' as the basis, though rarely informed how it is to be calculated. Presumably it is the average output which the firm expected to achieve, taking good and bad years together. Hence, if the oncost rates are calculated on this output and selling prices are based on the total cost including such oncost rates, the business would make the anticipated profit, provided the anticipated output could be sold at these prices. But if demand has changed there is no reason to assume that maximum profit will be made by charging a price to cover these fixed rates of oncost. Suppose a firm had anticipated a demand of 10,000 units per annum and had put down plant accordingly. Let us assume that the cost per unit, including overhead, on the basis of 10,000 units is £1; that demand, however, falls so that to market 10,000 we have to reduce price to 19s/95p per unit; that at £1 only 5,000 units will sell. In making its output decision it does not require to know that to cover overhead on 10,000 units it needs £1 per unit or that on 5,000 units it must get £2 per unit. What is required is information which in the given circumstances will help to fix prices or output at such a level as will maximize net profit. This is our entrepreneurial problem.

IV. PRICE-FIXING AND OUTPUT DECISIONS

For the purpose of exposition let us assume a factory turns out only one product—what cost information would be required in price fixing? If the market is perfectly competitive, the producer has no choice as to price, all that can be varied is the output and this can be expanded profitably until the last additional unit involves additional costs just less than the revenue added by that unit, that is to say, its market price. To take a simplified example, assume the selling price of an article in perfect competition is £2 10s 0d (£2.50) per unit, and the monthly costs are as shown in Table 4.1.

Table 4.1
Output in units	2,000	3,000	4,000
	£	£	£
Materials	2,000	3,000	3,900
Labour	1,000	1,500	2,250
Overhead:
variable	1,000	1,250	2,200
fixed	1,000	1,000	1,000
	5,000	6,750	9,350
Sales	5,000	7,500	10,000
Profit	nil	750	650

The most profitable output is clearly round about 3,000 units; it may be below or above this figure, and to find it exactly the costing department would have to study how each additional unit produced adds to the cost, and the management should stop increasing output at that point where the last additional unit involves an addition to total costs of £2 10s 0d (£2.50). In going beyond this point we should be throwing money away. Long-run selling policy might make it necessary to throw money away for short periods, but this should not blind us to the fact that short-run profits are not being maximized. Moreover, if the market is not absolutely competitive, then additional units of output sold would after a time add less than the price to total revenue, as prices would have to be dropped to dispose of the supply. In this case the output at which profit is a maximum is not that at which price equals additional cost, but that at which the additional revenue obtained from producing and selling one more unit of output is equal to the addition to total costs incurred in producing that additional unit.

For example, a firm might dispose of 3,000 units at £2 10s 0d (£2.50) each, but we can suppose that to market 3,100 the price would have to be dropped to £2 9s 0d (£2.45). Therefore, 3,000 units yield £7,500 and 3,100 units produce £7,595, hence the additional revenue from the increase is £95, and we have to see whether the additional cost resulting from the increase in output is greater or less than this.*54 Therefore the management will always be asking two questions, and bringing the two answers together. Firstly how will price changes affect total revenue? (This will depend on the elasticity of demand for the product.) Secondly what will be the additional outlay in producing extra units, or conversely, what will be saved by reducing output? Incidentally, additional costs are sometimes known as differential or marginal costs, and they are those costs which are variable.*55

It is useful to inquire whether there are any cost figures usually collected and examined which cannot affect policy. For example, the rent of the factory is likely to be the same if production goes on for one hour a day or twenty-four. In every business there are some expenses which are unalterable over wide ranges of output or over considerable periods of time. Hence there is no reason to study these unless major changes in output are being considered. For example, it might pay to incur the costs of moving the plant and machinery to smaller premises if output is to be reduced to half permanently. But, as suggested earlier, cost accounting normally deals with the ordinary run of production, special statistical investigations being made for major operations.

Textbooks are prone to emphasize the fact that cost accounting analyses past costs not future estimates, but they often do not make clear the fact that this data is useful only in so far as it is a guide to future costs; it is future variable cost which is important. Therefore cost accountants can ignore expenses which are completely unchangeable, e.g. the preliminary expenses of setting up a business. Sometimes, however, expenditure is composite—partly fixed, partly variable. For example, depreciation of machinery includes obsolescence, that is to say, the loss due to changes in values which the business cannot control. There may also be an element of physical wear and tear which continues whether machinery is used or not. It is therefore important to find out how and to what extent variations in use affect the total wastage. Similarly every other expense must be examined in order to establish the relationship between changes in cost and output variations.

Admittedly it is not possible to establish such a relationship with absolute accuracy, nor would it be possible to bring additional cost and additional revenue to complete equality without much expense. It is easily possible for the cost of accuracy to outweigh the advantages.

Many problems melt away if the subject is approached in the way which has been outlined. For example, from about 1890 there has been a stream of literature repeating ad nauseam the arguments for and against the inclusion of interest as a cost. All we really need to ask is: will the additional output involve the tying-up of capital which could be used or invested elsewhere? If so, the interest that the capital could earn elsewhere is a cost. On the other hand, interest on capital tied up in machinery is not important because the capital is sunk and could not be invested elsewhere. But if major changes, like the closing of a department are under consideration, interest on the scrap value of the machinery is a cost, because the money could be invested in alternative uses. Another problem concerns the price to adopt in charging out raw materials; one school claims that materials should be issued at original cost, while the other side champions 'replacement' cost. The 'original' cost supporters quarrel among themselves as to the way in which cost price is to be arrived at. Some use the 'first in, first out principle', others 'the average price of stock on hand'. Surely what we have to decide is the additional cost which the use of the material imposes on the business. This additional cost is the replacement price of the materials if they have to be replaced; but if it is not intended to replace them, then scrap value would be more appropriate.

Let us now consider a more complicated case; a factory turning out two products, A and B, each of which goes through two processes, the first of which is common to both products. So far as the costing of the first process is concerned the position is the same as it is in the 'one-product' factory. The management will require to know the cost of additional units at various levels of output. Usually in a costing system the total expenditure of the business would be divided between the three processes carried on, that is to say, rent, rates, insurance, administration, and so on, would be allocated on various arbitrary bases considered 'fair'. So far as these costs are fixed and unavoidable, it does not matter how you allocate them; as long as the nature of the business remains the same these costs go on. But it is true that space and other available services used for one process in the business could possibly be employed on the others, so that between the products there is a degree of variability, the cost of keeping one process going being equal to the opportunity of net gain by using the resources in the other processes. Starting from a position of equilibrium (i.e. one in which the net profits cannot be increased by changing the output or price of either or both products) let us assume the demand for product A increases. In deciding what changes to make, the effect on costs of changed output would have to be examined. The sales department must provide data showing estimated changes in the quantity of sales as prices are varied. The cost department must estimate the variation in costs which would be brought about by changes in output. Normally it would be necessary to consider only the additional costs in the first process (which is common), and the second process for product A. But other changes may be envisaged; for example, it may pay to expand the premises and put down fresh plant, rather than face increasing unit cost. If so, depreciation and interest on the additional capital is a variable cost to be taken into account. Or factors of production may be borrowed from the department making product B. Suppose there is excess space in product B department which can be transferred, then only the cost of alterations such as knocking down partitions need be considered, but if by cutting down its space product B department is involved in higher costs to produce the same output as before, then these additional costs must be added in as part of the cost of increasing the output of product A.

It may be protested that unless arbitrary allocations of departmental expenses are made it is impossible to see which department is paying best and should be expanded. This is untrue, as we test the profitability of increased output by examining marginal variations in cost and revenue. In other words, we compare increments to cost with increments to revenue, rather than totals or averages. Such an examination may show that it will pay to increase the output of one product at the expense of the other, and the only way to tell how far the change should go is to compare the additional revenue from one product less the reduction in revenue from the other with the additional costs incurred by the business as a whole as a result of the change. One cannot decide which product to increase, and by how much, merely by looking at aggregate periodic departmental accounts in which fixed costs have been allocated in some way. If we cannot use the information why prepare it?

The job costing of an engineering works is a much more complicated affair than the simple examples we have taken. Job accounts are prepared to show the materials, labour, other direct expenses and overhead incurred on each job, in order to show what profit each job has yielded, to provide data for future estimates and to control efficiency and prevent waste. No job should be taken unless it covers the variable expenses it incurs, except for such special purposes as maintaining a labour force, holding a trade connection or forcing out a rival—even in such cases it is important to know the cost of the policy adopted. This cost will be the difference between the variable expense on the job and the price received for it, assuming that the latter is less than the former. Details of variable cost should therefore be collected. But the greater part of oncost which is added to the job for costing purposes is fixed and goes on regardless of changes in output. Hence as a general rule any job yielding more than its own variable cost adds to the revenue of the concern and should be accepted*56 unless it is believed that by taking it a more profitable one will have to be refused later. Variable cost marks a minimum to price but actual quotations will depend on market conditions.

It is useful to add oncost to the job in the cost ledger? Will it help in policy? Let us examine the make up of, say, a machine-hour rate of oncost. Under this method of distributing overhead all the expenses of a department for a year are allocated to the machines in it—the latter are treated as production centres. Some 'normal' output is assumed, and on this basis the number of running hours for each machine is calculated. The cost of power, superintendence, heat, light and rent of the department are allocated to machines (e.g. power is often metered out, and heat, light and rent charged on floor space). To this cost is added depreciation and repair of the machine and sometimes interest locked up therein. The total cost divided by the number of machine-hours gives the hourly rate on the basis of 'normal' output (which incidentally does not usually mean that rate of output at which cost is a minimum). Each job is charged with oncost according to the number of hours for which it uses the machine. Now, looking backwards on finished jobs, can the management derive any help from examining these figures of total cost? The machine-hour rate hides the distinction between fixed and variable costs and tends to convey a false impression of variability. It does not tell you whether you did right in taking on the job. Looking forward, it is necessary only to estimate the additional cost which will be incurred in taking on work—this is the minimum to be accepted and if anything above this can be obtained then the job is profitable. In tendering for orders knowledge of market conditions will govern the bids, not estimates of total cost. There is no reason to assume that it will be any easier to guess the prices of competitors by calculating one's own 'normal' cost. In fact the oncost rates are likely to be more hindrance than help, because they contain in a confused mass both variable expenses and fixed costs. Our conception of the total cost will be no guide to the bids of competitors for an order, as these latter will depend on the state of the competitors' order books, and in any case methods of computing oncost and estimates of normality vary so much between accountant and accountant that it would often be dangerous to suppose that one's competitors have allowed roughly for the same oncost as oneself.

V. THE COSTING OF BY-PRODUCTS

Another very important problem dealt with in a most unsatisfactory way concerns the costing of by-products. There are several schools of thought among practitioners. For instance I think it is true to say that the American meat-packing industry regards dressed meat as the main product and all the other products such as hair, hides and wool are treated as by-products. The costs of handling the by-products are subtracted from the income derived from their sale. The net proceeds are then credited in the main manufacturing account, reducing the cost of meat accordingly. Thus all the profits are attributed to the main product. On the other hand, most firms in the oil-refining industry use the selling-prices of the products to determine the costs. For example, crude oil is split into five products. The account for that particular process is debited with the cost of crude oil together with the process costs and the total is divided among the five products in proportion to their market values. So by this method each product shows the same percentage of profit. A third method is to split the total in accordance with some chemical or other arbitrary formula, for example, on the basis of atomic weights.

In this connection there is a story told by an American economist, T. J. Kreps.*57 In a chemical works which was virtually controlled by a large bank the joint costs of a process were being allocated between the two products, caustic soda and chlorine. Owing to the method of allocation adopted fifty per cent to each product, the chlorine showed a loss. The absentee bankers wanted the chlorine foreman discharged, but the works manager, realizing that he would lose a good man, worked out a new cost allocation formula which was more favourable to chlorine; this product then showed a profit and the foreman kept his job. The new allocation was 56.73 to 43.27, and its pseudo-mathematical precision was the result of splitting costs in such a way that both products should show equal book profits. Obviously costing of this type is unsatisfactory and is no help in price policy or in controlling efficiency; the expense of allocation is money wasted. If the proportions in which two products come forward are absolutely fixed, then the joint costs of the process cannot be allocated between them, but if the proportions can be varied within limits and the variation alters the total cost, then it should be the job of the costing department to investigate changes in the cost arising out of changes in proportions, for to maximize net profit it will be necessary to watch price changes of raw materials and finished products and to vary the proportions of the two products to the point at which the added revenue from the last small variation is just balanced by the added cost.

VI. COST ACCOUNTING AS A TOOL OF EFFICIENCY

So far we have considered a few points arising out of the entrepreneurial problem, but the technical problem has not been discussed. We hear a great deal about costing as an instrument for producing efficiency and cutting down waste to a minimum. But I think we should bear in mind that excess capacity does not necessarily imply inefficiency. Much is made of the statement that costing shows you the cost of idle capacity. What the management does want to know is whether the output it has agreed to produce could in any way be turned out more cheaply. Can it combine its resources in such a way as to lower the total resources required? Is there any waste it can avoid, which is greater than the expense of avoiding it? Many records at present in general use are valuable from this point of view; for instance, perpetual inventory usually pays because it imposes a control over waste and theft and helps to insure that production is not held up for lack of raw materials. Moreover it provides records which enable the management to reduce to a minimum the capital tied up and therefore prevents loss of interest and wastage due to obsolescence. Plant ledgers are useful because they afford a convenient way of collecting information as to the performance of machines and facilitate the calculation of depreciation.

It is not proposed to discuss the many ways of increasing efficiency which the textbook writers catalogue. Modern works on accounting have tended to give too much space to this aspect of management. Given that the rate of output is a settled question, then so long as relative prices of the factors of production remain unchanged, the efficiency question is not one for the cost department, it is just a matter of vigilance on the part of the works manager. But as soon as the proportionate prices of resource change it is the job of the cost office to see whether the combination of resources can be altered, in order to prevent a rise or bring about a fall in the cost of production by increasing the use of the relatively cheaper resources and decreasing the use of the relatively dearer ones.

One development of cost accounting which has received much publicity in connection with efficiency is known as standard costing. Studies are carried out for the purpose of finding either the cost to be expected under normal conditions or under ideal conditions at different levels of output and these are used as foot rules to measure actual performance; in this way the standards are to be a sort of incentive to greater effort. Some of the systems are exceedingly complicated and the standards are incorporated into the double-entry book-keeping—the hallmark of respectability. If the management can control slackness and create incentive by using these figures, then they may be justified, provided there is no cheaper way of doing the same job. For example one might work out that rate of output of a machine which could be produced at lowest cost. After calculation of the optimum capacity of each department from such computations, it might be possible to estimate the flow of output through the plant which would result in lowest unit costs. This would represent technical but not of course economic perfection (unless competition were perfect). In any case studies of the effect of different rates of flow of production would enable the management so to arrange its output within the budget period as to achieve the minimum cost possible for that output.

VII. SOME CONCLUSIONS

I believe that cost accountants have spent too much effort in trying to arrive at total cost by building up complicated and delicate oncost structures which depend on arbitrary assumptions. But on the other hand in some industries long-period analysis can be helpful to the management and its estimating department. Although I consider it the cost accountant's main job to inform the management regarding the minimum at which additional work can be taken, this minimum will vary according to the extent to which capacity is being used, and will sometimes include capital costs. For example, if a firm is already working at full capacity, then any further output involves additional capital outlay and the revenue obtainable from the additional turnover must cover the amortization of the new capital outlay in order to be worth accepting. This, however, is not all; an engineering firm, for instance, has to estimate and tender for work. It does not know the estimates of other firms; the only information of which it is certain is its own minimum price, which will be different in periods of normal activity, in times of boom and in times of slackness; this minimum will in each case be the additional cost. But it should also know that unless it gets prices including overhead it will not be able to replace its fixed assets. This does not mean that it ought to charge these prices—to do so in some conditions would put it out of the market altogether; but the management should see that the firm is coming to an end. The cost department should say definitely at what figure a job is worth handling and possibly how much we ought to get if we are not to close down when our fixed equipment wears out. To do this overhead costs must be allocated over jobs in the least arbitrary manner possible. There is no time to go into this wide question, but I would like to emphasize that it is future costs we have to deal with not past ones. One has to provide in overhead the cost of replacing assets; the original cost is of no importance; the past is irrevocable. It is, however, of interest to a firm to know whether it is getting enough out of contracts to cover replacements costs. It may be argued, and with some point, that a detailed analysis of overhead is not worth while for this purpose and that the annual accounts will give sufficient warning. But in any case, if the future of costing lies principally in statistical examination of marginal variations, then it may be doubted whether it can be fitted into the framework of double-entry book-keeping.

Within the time remaining this evening it is impossible to make concrete suggestions as to how the analysis I have attempted to describe should be applied to individual cases, but I think that for each department the accountant should prepare, and continuously revise, schedules showing the additional cost of additional output. For this purpose, the cost accountant would need to be provided with details relating to market prices of materials; he would require a wealth of analysis concerning the expenditure of the business, and engineering data showing how, for example, the rate of wear and tear of machinery is affected by use. Overtime rates, fatigue studies and so on should be part of his stock-in-trade. He should, for example, be able to compute the additional cost of running nine hours per day instead of eight, or the cost of increasing the speed of machinery. He should be able to estimate the cost of an increase of output over and above the budget figure. In those cases where demand fluctuates it should be possible to decide on the cost accountant's evidence how far it is worth while to make for stock in the valley periods. Again the accountant's figures should show whether in a depression a smaller loss is made by selling at a known margin below variable cost than by closing down for a time.

Thus most of the cost accountant's data will come forward in the form of statistical statements, in the preparation of which little help can be derived from a 'tied-in' double-entry system. Of course certain information in individual cases may be wanted so often that it is cheaper to collect it continuously even though it may at times be useless, but this course can be carried too far. Many firms order the continuous collection of data, much of which is required only at infrequent intervals and some of which is never required at all. The cost of this continuous collection must be compared with the cost of a separate investigation each time the data is required, bearing in mind the fact that information to be of service, may be required at very short notice.

Although some criticisms of present methods of costing have been suggested, attempts to allocate fixed costs may be justified on grounds quite unconnected with the problems we have discussed. If time permitted it would be interesting to discuss the growth of uniform systems of costing advocated by many trade associations on the grounds that ignorance of manufacturing and selling costs induces unpleasant price competition. According to T. H. Sanders, Professor of Accounting at Harvard:

Some industries are especially characterized by the presence of large numbers of small manufacturers who are likely to pay little attention to costs, and as a result jeopardize the success of everybody in the business. The larger manufacturers have a genuine dread of competition originating in such sources, and one of their most effective means of combating it has been the development of cost-keeping methods which would tend to remove the prevailing ignorance.*58

Thus systems of uniform costing are said to aim at preventing price competition due to ignorance. It is, however, doubtful whether a large proportion of goods are sold below total cost merely because the manufacturers know no better; it is more likely that most action of this sort is deliberate. But either way, if some manufacturers choose to make a present of part of their output, this does not reduce the price which other manufacturers are able to charge for their goods. As Professor Plant suggested to the writer, hospitals cannot usually buy bread more cheaply merely because some bakers on occasion give them free loaves. It may be true, however, that a uniform costing system can be used for fixing prices in order to bring about tacit monopoly. Whatever the long-run effect of this, e.g. in attracting new firms to the industry, it is certainly true that suppression of price competition usually leads to competition in quality or service, and the small firm should consider carefully its position in this respect, when opposed to large undertakings.

Sometimes in order to obtain contracts 'window dressing' is required, e.g. the 'cost-plus' basis may be in use or it may be necessary to satisfy a purchaser that the prices charged are 'fair' or 'reasonable'. For bargaining with the unsophisticated it may be very useful to point to fixed costs divided over normal output (output being calculated on the low side) and to say 'This is what it costs us'. This may be partly cost accounting and partly salesmanship. But, of course, in so far as buyers are dependent on the continued existence of supplies from certain firms they must be prepared to pay long-period costs, including sufficient to replace assets as they wear out.

It may be pointed out that finished stock and work in progress have to be valued for balance-sheet purpose and that current methods of costing are useful for this. The basis generally adopted for finished stock is 'cost or replacement value, whichever is the lower', and this method is accepted by the Inland Revenue for income tax purposes. These valuations generally include a proportion of oncost, the amount of which depends on the views of the accountants concerned. One of the practical objections to treating interest as a cost has been that to do so would 'anticipate profit' on unsold stock and work in progress and, incidentally, income tax would be attracted sooner than need be. Some firms in valuing work in progress exclude all oncost to be on the safe side. But the Revenue authorities object to this on the ground that tax collection is delayed thereby. It might be claimed by the tax payer that the value of unfinished goods is so problematic that nothing should be added to the variable cost, but on the other hand the Revenue authorities might well contend that to a going concern the value of the work in progress is equal to the net selling price of finished stock, less the additional costs required to complete the work in progress. In practice compromise is reached by adding an arbitrary proportion of fixed costs. The whole procedure is unsatisfactory and the principles of valuation require re-examination by accountants; it is unlikely that a costing system allocating fixed cost is justified on these grounds alone.

If I may be allowed to summarize my views, I should say that we can distinguish three lines of approach to the costing problem.

Firstly, information is necessary to enable the most profitable output to be decided. This depends on marginal revenue and marginal cost. We have called this the entrepreneurial problem.

Secondly, information is necessary to ensure that the proposed output is produced at the lowest cost possible for that output having regard to the facilities available. This is the technical problem of combining factors of production and avoiding waste.

Lastly, there is the book-keeping problem of deciding how best to deal with the first two questions having in mind that collecting information costs money.