Accounting For Capital Revaluations
The method which we here advocate for the measurement of the proportion of capital resources which annually enter the social product is a method which Dr Fabricant of the National Bureau of Economic Research employs for the determination of capital consumption in the United States from 1919 to 1935.7 Though ostensibly his aim is the measurement of the difference between gross and net national income, it is evident that this difference equals the total capital change other than gross capital formation. More important is that for Dr Fabricant, being
thus interested in one type of capital change, the determination of capital consumption is only part of his task. For only capital consumption, the amount of capital used up in production, and capital adjustment, the capital changes due to causes external to the production of output, together make up total capital change other than gross capital formation, i.e., the capital change he seeks to determine. The measure of capital consumption he finds in annual charges to income account (depreciation on fixed capital, depletion of wasting assets, plus some minor items like provisions for accidental loss and maintenance and repair expenditure), the measure of capital adjustment in (intermittent) charges to capital account (retirements and abandonments and revaluations of capital assets8). In so far therefore as the object of his investigation coincides with ours, i.e. fixed capital used up in production, he virtually employs the method we advocate.
But it may be doubted whether in actual accounting practice the distinction between charges to income account and charges to capital account can be regarded as symmetrical to Dr Fabricant's distinction between 'internal' and 'external' capital change. If there is a parallelism, it is at least not a very close one. On the other hand, obsolescence, certainly the result of 'external' factors, is usually provided for in depreciation in so far as it can be foreseen.9 On the other hand, there is the case, well known from the history of all inflations, where depreciation allowances, because based on original rather than reproduction cost, habitually understate the amount of capital used up in production, with the result that sooner or later a charge to capital account has to be made. It seems to us, therefore, that the accounting distinction between charges to income and to capital account is better regarded as reflecting the economic contrast between foreseen and unforeseen changes.10 Nevertheless, it still remains true that we have to exclude charges on capital account from our computation of capital used up in production. Our next task consists in explaining why this is so.
In a world of perfect foresight depreciation allowances, in so far as they cover 'normal' obsolescence as well as ordinary wear-and-tear, will, as a rule, suffice to ensure capital replacement on a scale commensurate with the permanent flowing of a constant income stream. But in a world of imperfect foresight this need not be so, and the aggregate of depreciation allowances over a number of years may, for familiar reasons, either exceed or fall short of this standard; hence the recurrent necessity of revaluation of capital assets. The frequency of such revaluation offers strong evidence that in a world of change the 'maintenance of capital' is not a practical possibility. Adopting Swedish terminology we might say that depreciation allowances are ex ante estimates of the amount of capital used up, whereas the subsequent correction of these estimates by a charge to capital account determines this amount ex post.
For the purpose of the measurement of capital changes during a sequence of years there are three alternative methods of dealing with these intermittent revaluations. Each of them has its merits according to the purpose in hand and the aspect of the problem we wish to stress. There is first Dr Fabricant's own method which consists in simply adding up charges on income and on capital account. Since he is interested in total capital change other than gross capital formation over the period 1919 to 1935, this is no doubt an entirely legitimate procedure. A second method, more, we believe, in accordance with the essential meaning of period analysis, would be to include or exclude capital adjustment from our computation according to the length of the period chosen. For instance, when computing the quantity of capital entering annual output we might well confine ourselves to depreciation and disregard revaluation, while we could not afford to disregard it where, like Dr Fabricant, we had to deal with a period of more than fifteen years. In adopting this method we would do more than merely draw a conventional distinction between 'long' and 'short' periods and the forces operating in them; we would, that is, implicitly admit that there are forces making for capital change which are quite unrelated to the productive contribution of capital. It is on a recognition of this fact and its implications that the third method, the omission of capital adjustment from the computation of capital, is based.
We learned above that the difference between foreseen and unforeseen changes provides the economic rationale for the accounting distinction between charges to income account and to capital account. As in general the effect of unforeseen changes on capital value cannot be apportioned to individual income periods, it is only logical to charge it to capital account. Yet this does not equally apply to all types of unforeseen change, and in fact there is a further important distinction which is germane to our argument. There are, on the one hand, changes which, although unforeseen at the moment at which the production plan was made, do not, when they occur, altogether upset the plan, do not necessitate any important modification of the pattern of resource use as chalked out in the plan (there may and always will be minor technical adjustments), and which might very well find their financial expression in charges to income account if accounting methods were more elastic than they actually are. A good instance of this, as was mentioned above, is provided by the history of inflations, in particular their earlier stages: Where in such a situation depreciation methods are based, as they traditionally are, on original rather than reproduction cost, sooner or later a charge to capital account will have to be made, the necessity for which, however, arises not from the nature of the case but from the nature of bad accounting. Where this happens we are, of course, unable to accept depreciation allowances as compiled from actual business records at their face value, and in a moment we shall have to discuss methods of their rational reinterpretation. But there is another type of unforeseen change, arising, for instance, from technical progress or changes in the social environment in general, which upsets the whole production plan and makes it impossible to adhere to the pattern of resource use as originally chalked out in it. In the former case it was still possible, after the change had occurred, to allocate its effects to single income periods. In the case of inflation all we have to do is to allow for the higher cost of replacement goods in the depreciation allowances for subsequent years. But in the latter case no such allocation is feasible, for here the very occurrence of the change marks at the same time the failure, and consequently the end, of the production plan. Now an entirely new plan has to be drawn up, and the failure of the first finds its appropriate expression in a charge to capital account.
We contend that in the computation of the amount of capital which gives rise to a certain stream of services, and hence indirectly to a given output stream, charges to capital account in so far as they reflect the latter type of unforeseen change must be disregarded. In not doing so we would obviate the very reasons for which we adopted period analysis, for our chief reason was our recognition of the fact that in a world of change the measurement of capital over periods so long as to involve alterations in the mode of its use is logically impossible and, if nevertheless persisted in, bound to lead to absurd results. Hence, in applying period analysis we have to select our period so as to make it co-extend with the carrying out of a coherent production plan. But revaluation, as we saw, usually marks the failure of a production plan. It follows that our period has to be selected in such a way as to have it ending before such necessity arises.
However, as we pointed out, in reality asset revaluation often marks not the failure of a production plan but the failure of accountants to understand the nature of the processes the effects of which they are about to assess. Frequently charges to capital account are made where charges to income account would be quite as feasible. It follows that if in our computation of capital resources we disregarded these charges to capital account like all others, we should be disregarding something which really ought to be included in depreciation. Hence, if we wish to omit all charges to capital account from our computation, we must somehow modify depreciation data as compiled from business records in order to make them cover these cases.
Here we are extremely fortunate in being able simply to follow Dr Fabricant, who has worked out most ingenious methods in order to deal with this problem.11 By making judicious use of an inquiry by the New York Stock Exchange of companies applying for the listing of their securities, he first establishes the principles to which actually the majority of American companies adhere in their depreciation practice.12 As was to be expected, it is found that the principles underlying their practice are very different from what they would have to be if the data compiled from balance sheets were to be of immediate use for our purposes. Still, the result of the inquiry reveals that, however illogical these methods may be from the economist's point of view, there are certain broad underlying principles to which the large majority of company accountants adhere and which are not incapable of rational reinterpretation.13 The two main defects of depreciation data are found to be that they ignore variations in the degree of utilization, and that they are largely based on original rather than reproduction cost. But Dr Fabricant shows convincingly that these are by no means insurmountable obstacles. If depreciation allowances are to reflect the amount of capital resources responsible for the output of a period, they will have to be on a 'service-output' basis,14 i.e. they must vary with variations in output and the degree of utilization. But in practice (this is one of the principles established by Dr Fabricant) depreciation is usually on a 'straight line' basis, i.e. annual depreciation is computed by dividing the original cost of capital equipment through the prospective number of years of its economic life, and irrespective of the variations in output.15 We therefore have to construct an index of output by means of which we can convert our data from a 'straight line' to a 'serviceoutput' basis.16 The same applies to replacement cost. Here we have to find the age-distribution of industrial equipment,17 and then to construct a price index for equipment goods.18 This done we are able to convert our data from an original cost to a reproduction cost basis.
Finally we have to show why the method here proposed is not open to the objections which in the first part of this chapter we raised to Mr Clark's. After all, it might be said, our data as well as his are finally derived from book values of capital assets, as depreciation allowances computed on a 'straight line' basis are, of course, merely a function of book values. We made our statistical raw material pass through a processing stage, but why could not Mr Clark's data be subjected to similar treatment?
The answer is that by disregarding capital adjustment we contrived to eliminate those capital changes which are due to unforeseen events entailing discontinuity in the mode of use of capital resources, and hence irrelevant to the computation of the amount of capital used up in production. Our main charge against Mr Clark was that the object of his statistical measurement was not identical with the aim of his investigation, viz., the non-permanent sources of productive services. On the one hand, his data were seen to be affected by changes in book values of capital assets which did not reflect concomitant changes in the productive contributions of these resources. On the other hand, his data did not reflect changes, like variations in the degree of utilization, without taking account of which it is impossible to determine the productivity of any factor. The former problem we endeavoured to overcome by ignoring charges to capital account and by the choice of an appropriate time period, the latter by correcting our original data for changes in prices and output. Instead of measuring a stock of resources capable of giving rise to service streams of various length and shape we prefer to measure the depletion of the stock consequent upon the flowing of each service stream.
One objection we hope will not be raised against us, viz., that by basing our computation of capital on subjective estimates we are infringing upon the objectivity of measurement. Such argument would incidentally raise a fundamental issue in the methodology of social science, and it would be possible to meet it on a broader plane by discussing the meaning of objectivity in social life. In the present context we may forestall it by pointing out that in the field of capital, where the physical attributes of the components of this heterogeneous mass cannot serve as criteria of classification, all measurement must necessarily proceed in terms of value relationships, and all economic values are ultimately derived from subjective estimates. We readily agree that where a market is sufficiently large, generally accessible, and continuous over time, it serves to co-ordinate a large number of subjective estimates and thus may impart a moment of (social) objectivity to value relations based on prices formed on it. But it can hardly be said that the second-hand market for industrial equipment, which would be the proper place for the determination of the value of capital goods which have been in use, satisfies these requirements, and that its valuations are superior to intra-enterprise valuation. On the other hand, as Dr Fabricant has found, business accountants in their depreciation practice as a rule follow a general but very distinct pattern of behaviour. If there is such a thing as institutional objectivation of subjective attitudes, it applies to modern business accounting certainly no less than to market values.
We must here confine ourselves to this brief and very sketchy outline of an approach to a problem the intricacy of which has hitherto perhaps not always been fully recognized. We are painfully conscious of its many shortcomings, like the snares implicit in the construction of a price index of capital goods under conditions of technical progress or in disregarding intersectional output variations, but none the less we venture to put it forward as a contribution to the discussion.
Post a comment