102

A TREATISE ON MONEY

BK. II

And the comparison of purchasing power for a middle-class parent before and after the war may also bebetter made by general impression, because so sub-stantial a part of the expenditure hi question—asrent, servants, education and travelling—is of akind which inevitably escapes inclusion in an index-number.

It is in such cases as these that the direct methodmay yield a result materially different from the in-direct method and yet be nearer the truth. When,on the other hand, we are dealing with standardisedexpenditure which has not much changed in character,as for example with that of the working classes nowand five years ago, the indirect method will probablybe much more accurate.

Whether it is a merit or a defect of comparison bygeneral impression that it will usually make full allow-ance for conventional expenditure which yields littleutility but is required by local custom if one is to avoiddisutility, raises a subtle question of just what onemeans by purchasing power which I will not allow todetain me.

B. The Indirect Method of comparing Pricesof Equivalent Composite Commodities

This is the usual method. In the form in whichthe problem of establishing index-numbers of pur-chasing power is generally set, we know, at best, theprices of different commodities and the distributionof expenditure between them, and we have little orno information about what pairs of persons are“ similar ”, wherewith to supplement these figures.

We generally do know, however, that some partof the expenditure—and sometimes a larger part—isin both positions identical in character and in powerto yield satisfaction to pairs of similar consumers. Letus represent that part of the composite commodities