64
(a) that in the time available it had not been
possible properly to examine and verify the returns of income and expenditure submitted; a need which had been the first lesson learnt from the University's earlier experiment;
(b)
that we had grave doubts about the justificat- ion of the system of weights used;
(0) that the effect of the omission of all
consideration of the fact or of rent in the second survey had the effect of producing an cxaggeration in the deduction on the cost of living to which the index lcd: a form of error, the danger of which was clearly indicated by the Statistical Officer and for which he suggostod in a later note apponded to his
Our caution report, a tentative correction.
in using the results of the surveys is justified by the opinion of one of our most valuable witnesses who wrote:
The
"The Report (S.0.3/47) on an "Inquiry into the Cost of Living of Asiatic and Portuguese Clerical and Skilled Technical Employees", prepared by the Statistical Branch of the Colonial Secretariat, provided useful data in regard to the range of certain retail prices in January-llard: 1939 and January-March 1947. For reasons given in the Report no attempt was made by the Statistical Branch to calculate a Cost of Living Index Figure. A general Retail Price Index Figure was, however, drawn up in the Report on the basis of a carefully explained procedure, rosulting in a figure of 647 (January-March 1939 100) for the period January-March 1917. The corresponding Retail Food Pricos Index
Thoso figure was calculated at 781.5. estimates, if near the mark, had, clearly, great significance for any comparison between the value of purchasing power of money in 1947 and its valuc in 1939, for the type of households comprised in the survey. extimated Index Figures for 1947 were considerably higher than had boon expected. An examination of the weighting system adopt- ed suggested the possibility of some duplicat- ion of omphasis. The large weight given to food in the Goncal Price Index for example, is itself the reflection of higher prices; within the food index, the largo weight a allotted to Free Market Rice (22.06) is also due to the fact that the price is much highor than in 1939. If then these weights are used to multiply the pe.contage increases in price, thoro seoms a danger of double counting.
While that is a matter for statisticians, it would scem that an inquiry into the quantities of each item bought, on a monthly average per household, might provide a useful check; since then it would be possible to compare the actual cost of these quantities in January-March 1947 with what they would have cost at January-March
72