COST OF LIVING INDEXES COMPARED 119
samples, so as to represent total consumption habits, are
necessarily matters on which there is great leeway for the
judgment of the investigator, and which cannot but affect the
results. If it 1s understood that the Bureau of Labor Statis-
tics clothing budget is based on war-time consumption and
includes a quantity of yard goods and that the National
Industrial Conference Board budget is based on pre-war
consumption and contains no yard goods, many if not most
of the differences between the two clothing series can be
explained; what remains is due largely to matters of taste
and judgment.!
Fuel and Light: For the fuel and light item, sometimes the
"This position is at variance with the opinions expressed in Carr, Journal of the
American Statistical Association, December, 1924, op. cit, pp. 503-506, in which,
however, no mention is made of the difference in the relative quantity of yard
goods and made up garments included, the possible difference between 1914 and
1918 standards of samples, the effect of the reduction in the number of articles in-
cluded by the Bureau or the increase in the index which followed the Bureau’s
adoption of a system of weights in 1920. It is there suggested that the difference
between the Board’s figures and those of the Bureau may be due to some of the
following, in each instance assuming at the outset that the Bureau’s procedure
is the better: (1) the clothing budgets of the Board are not complete, but changes
in the cost of several items are assumed to have been in the same proportion as
changes in the cost of certain other very similar items. While the Board was
criticized for possible omissions of details by which the effect of this could be
checked, the oo was not noted that the Bureau of Labor Statistics in no case
publishes the details of price changes for the separate items in its clothing index,
y means of which the latter can be checked. (2) Prices of children’s clothing are
not collected by the Board, and there is a difference in the trend of children’s
clothing prices and those of adults, “Children’s clothing does not decline neatly so
much as does adults’ clothing.” The figures given show that, in the average for
the ten cities computed, the difference between the indexes for adults’ clothing
and children’s clothing was 4 points on seven dates and 3 points on seven dates.
Assuming, for the sake of argument, in the absence of the exact figures of the
Bureau, which are not published, that the total cost of the adults’ clothing budget
would be twice that of the children’s, a conservative estimate, the total clothing
index falls somewhere between the two and is more nearly like that for adults than
for children. The difference between the clothing index for adults alone and for
children and adults combined would not be more than one point, certainly not a
serious consideration in anything so entirely a matter of averages. (3) The Bureau's
use of special agents is also considered more “accurate” than the Board's use of
questionnaires, and (4) some difference between the two clothing price series is
attributed to the fact that the number of the Board's reports varies, while that of
the Bureau remains constant, and that as substitutions are made from time to time
the effect of dropping out and adding cities is to change the trend of the index
because price trends are different in different parts of the country. The illustration
given, however, indicates that, grouping the 32 cities from which the Bureau collects
quotations into four geographical areas, and computing the change in clothing
prices from June, 1920 to December, 1920, the average declines were 10.3% in the
East; 10.79, in the Middle West; 10.5% in the South, and 8% in the Far West.
This seems pretty clearly to bear out the Board’s experience that geographical
location as such does not affect clothing price trends, but that the character of the
stores reporting, and the maintenance of their comparability are far more important.