I mentioned earlier the desirability of introducing statistical design, as far as possible, into the procedure for collecting economic information in order to obtain the information in the most efficient way, to avoid bias and incompleteness, and to enable a measure of the reliability of the resulting estimates to be made. The field of social accounting is one in which a method of obtaining information which possesses these advantages is particularly desirable since here we are dealing with a system of empirical constructs which are related to one another. Moreover, measures of many of the component elements can only be made with considerable difficulty and uncertainty from information which already exists.

In attempting to demonstrate the isomorphism between the three types of account in a basic accounting structure and the three fundamental forms of economic activity I hope to have shown that it is possible in principle to collect the information required for social accounting investigations on an accounting basis. It is possible to draw up an accounting system which will reflect the kind of distinction in which we are interested as economists and it would be possible to use this system, appropriately modified and elaborated, for different parts of the economy as a basis on which to collect social accounting information. If this be granted then the difficulties we have to face are of a practical nature and the more important of them may be described as follows.

Having set up an accounting structure which makes use of the definitions and distinctions best suited for economic analysis, we have to see that this information can be obtained in practice from available records. This problem arises in a sense whatever method of collecting information is adopted, but it has to be faced explicitly if we start with a definite demand for information of a particular kind rather than with an acceptance of primary data as they already exist.

As far as business enterprises are concerned it can be seen that the accounting structure desired is not very different from the accounting structure in actual use. On many points of detail accounting conventions are satisfactory from the standpoint of an economist, but there are two kinds of problem which deserve special mention.