Published online by Cambridge University Press: 26 March 2020
Wherever possible, the monthly and quarterly statistics used in the Economic Review have been adjusted to eliminate seasonal variations. This note describes the methods of seasonal adjustment and discusses how far the adjusted series can help the interpretation of current developments. The last section of the note describes an analysis of imports which, although it did not yield a satisfactory set of seasonal corrections, may be of help to others.
(1) The limits were calculated at the 90 per cent level of confidence from the t-distribution for the appropriate length of the observed series. Detailed tables showing the standard deviations and the confidence limits for each quarterly correction will be supplied on request.
(2) It should be noted that this procedure gives rise to difficulties too: if there were a measure of the reliability of corrections to the change in a series between, say, the first and second quarters of the year and a similar measure for the change between the second and third quarters, there would still be no direct measure of the reliability of the change in the corrected series between, say, the first and third quarters. Moreover comparison is most often made not between two single quarters but between the latest quarter and a series of earlier quarters.