46 BULLETIN 1440, IT. S. DEPARTMENT OF AGRICULTURE 
depression as does the average of the commodities represented in the 
Bureau of Labor wholesale price index (17); but that in periods of 
exceptional business activity hog prices are affected to a greater 
extent than are the prices of all these commodities. 
The curve for steer prices shows that the sensitiveness of hog prices 
to steer prices varies with the price of steers, an increase in steer 
prices from "$5.75 to $6.25 increasing the hog demand index nearly 
4 per cent, whereas an increase from $6.25 to $6.75 increased it less 
than half as much. Beyond $7.50 further changes in beef prices 
had no marked effect on hog prices, if anything actually tending to 
lower them. This offers grounds for some interesting speculations 
as to the precise relations between substitute products; but it should 
be borne in mind that with the rather low total correlation (P = 0.81 
meaning that only about 66 per cent of total related factors have been 
measured, and that 34 per cent still remain unaccounted for) not 
much significance can be attached to minor changes in the curves. 
This applies also to some of the irregularities in the other curves — 
they may be due to chance fluctuation or to chance relations to 
unmeasured forces, rather than to the factor being directly measured. 
The curve for foreign demand is about as would be expected on 
a priori grounds, ignoring the double inflection, which may be due to 
the same random forces just mentioned. It is interesting, however, to 
compare this curve with the curve of export demand. (Fig. 27.) 
The " mirror" similarity suggests , that the shape of this curve may 
have been dependent upon the shape of the export-demand curve, 
and that better results might have been secured by obtaining the 
foreign-demand curve by some different method. 
FORECASTING PRICES IN TERMS OF DEVIATION FROM TREND 
The preliminary study on forecasting prices was made by testing 
the correlation between the independent factors and the dependent 
factors at various lags. 
The factors considered were : 
A. Corn-hog differential. 
B. Average live weight of hogs at Chicago. 
C. Index of prices of industrial stocks. 
C Harvard price index of business cycles. 
D. Price of corn. 
E. Price of hogs. 
The corn-hog differential was computed by multiplying the price 
of corn by 11.42 and subtracting the result from the price of hogs. 
The average weight of hogs was adjusted to remove average sea- 
sonal variation and then expressed as a percentage of the 39-month 
moving average, to eliminate the downward trend in weights. 
The price of industrial stocks was expressed in percentage of its 
moving average for a 45-months' period. This moving average 
showed an upward trend almost identical with the trend hi the 
general price level, so for later work an index of the price of indus- 
trial stocks was obtained by dividing the prices of the stocks by the 
Bureau of Labor all-commodity index of wholesale prices (17). 
The Harvard Price Index of business cycles, which is published in 
terms of percentage deviation from trend, was also used as a factor 
in the preliminary calculations (19). 
