BUREAU OF LABOR STATISTICS INDEX 69
Thus, a minute description of consumption habits and cur-
rent expenditures was obtained. From an analysis of these,
the conclusion was reached that among wage earners and
low and medium salaried families, annual expenditures were
distributed as follows: for food, 38.2%; housing, 13.49;
clothing, 16.6%; fuel and light, 5.3%; furniture and furnish-
ings, 5.1%; miscellaneous, 21.39%,. These differed, of course,
from place to place and with the size of the family and the
amount of the income; but for all families, everywhere, it
was found that out of every $1,000 spent, an average of
$382 went for food; $134 for housing; $166 for clothing;
$53 for fuel and light; $51 for furniture and furnishings;
$213 for miscellaneous items.
Although the basic budget on which the Bureau has esti-
mated increases in the costof living since 1913 is, to all intents
and purposes, based on a study of expenditures in 1918 and
1919, this is not true of all items, and although the trend line
produced by the weights thus constructed is not out of line
with other trend lines covering the same period,? it is a matter
of interest that these differences occur.
Food
For example, since January, 1921, the retail food price
series of the United States Bureau of Labor Statistics has
been based on average prices of 43 articles obtained in 51
cities, and weighted according to consumption in those cities
In 1918 and 1919. Prior to that date, however, a varying
number of articles had been used, beginning with 15 in 1913
and numbering 22 from 1915 through 1920. The consump-
tion weights for the earlier series were based on an investi-
gation made in 1900-1902. Inasmuch as these were two
distinct series, linked together in 1921, their make-up will
be discussed separately.
In 1900-1902 the United States Bureau of Labor Statistics?
4 United States, Bureau of Labor Statistics, Bulletin No. 357, op. ¢it., p. 466. The
total of the separate items is actually only $999.
2 The greater increase shown by the figures of the Bureau of Labor Statistics
than by other indexes of the cost of living is apparently due not so much to the
changes in weighting or to the weightings themselves, as to other factors. See
Chapter V of this volume.
® Then the United States Bureau of Labor.