COST OF LIVING INDEXES COMPARED 109
To test out this theory Table 6 has been constructed.
This gives the Bureau's index number series for each item
separately and for all combined, on the basis of the 1918
weights—that is, the official number. It also gives, in the
last column, a new series for the total cost of living, based
on the same index numbers for the major items, but com-
bined on the basis of 1913 weights. In other wotds, from
the known distribution of expenditures in 1918 and known
increases in prices between 1913 and 1918, the distribution of
expenditures in 1913, the base period, has been computed!
and substituted in the series for the official weights based on
the 1918 distribution. The difference between the two series
1s seen to have increased as the prices of the different items
advanced in varying proportions. At the peak, when the cost
of clothing had increased 187.5%, and that of housing only
34.9%, the maximum difference, 9 points, occurred. Where,
however, increases in the cost of the major items were more
nearly uniform and none of them so tremendous, the two
series were identical, or nearly so. In other words, at the
peak of prices, with very uneven price increases recorded for
the different items, the advance in the total cost of living
obtained by using wartime weights was greater than would
have been the case had weights been used based on pre-war
consumption. During most of the period of the series, the
difference due to the effect of the weights was negligible.
This test assumes, however, the consumption of the same
goods and services in 1913 as in 1918. In other words, it
changes the series only to eliminate the effect of price in-
Creases on expenditures. To convert the Bureau of Labor
Statistics series to a pre-war basis as regards consumption
habits, pre-war weights based on these habits should be
used. For this purpose, the increases in cost of the separate
Items since 1913, as found by the Bureau of Labor Statistics,
have been multiplied by the weights used in the National
Industrial Conference Board series for each date on which the
Bureau has figures. These weights are based on pre-war
consumption. The results are shown on the next page.
! The 1913 weights were determined by using the geometric reversal; the
arithmetic produces a total in excess of 100.
*See p. 32 of this volume.