44 BULLETIN 1440, U. S. DEPARTMENT OF AGRICULTURE 
the multiple coefficient to P = 0.81. Substituting the functional 
values of the variables for the variables themselves, 20 it was possible 
to measure the relative importance of each variable by the coefficient 
of determination. The results were as follows: 
Determination of demand index by — 
Per cent 
X4 Storage stocks 18. 5 
X 5 Business cycles 3. 1 
X 7 Price of steers 2. 8 
X 8 European demand 43. 3 
X 9 Trend in time 1. 9 
Total determination (R 2 ) 65. 8 
The time factor was included merely to eliminate any trend which 
failed to be eliminated in the previous computation of the index of 
demand. This proved to be hardly necessary, as the net regression of 
the index on time was practically zero, and it was of but little signi- 
ficance as a determining factor. Apparently the growth of demand in- 
cluded in making up the demand curve was about correct. 
The statements made in the section on "The demand for hogs' ' 
as to the relative importance of each factor were based upon the 
determination coefficients just listed. It is evident that fluctuations 
in industrial activity, in so far as they are reflected in the particular 
index used, are relatively unimportant in affecting the demand for 
hogs. It must be admitted, however, that this includes stocks in 
storage as a demand factor, when this really is one resultant of the sup- 
ply side of the equation. If the demand curve were determined 
directly from price and quantity consumed, and variations in de- 
mand measured from that base, the variations in business activity 
would probably be shown as a much more important factor affecting 
variations in demand. In addition, the demand index was de- 
termined from a price "deflated" for price level. As shown on 
page 43, price level changes accounted for 18 per cent of hog price 
changes. Some part of the changes in the price index is due to the 
business cycle ; so a portion of the variation in hog prices here ascribed 
to price level changes really is due to the changes in demand ac- 
companying changes in the business activity. 
So many assumptions are involved in the train of calculations 
from the computation of the demand index through this analysis of 
reasons for its variation that the particular net regression curves 
obtained are not thought to be of any great significance. They are 
shown here, however, that the record may be complete (fig. 29). 
When stocks were very low, demand increased as stocks began to 
be larger. In this case probably both changes reflect restored con- 
fidence. Then for a time there was only a slight relation. But 
when stocks became much heavier than 250 million pounds (in 
terms of normal season stocks) there was a precipitate drop in 
demand, a further addition of 50 million pounds to storage decreasing 
demand by 4 per cent. 
The curve for the effect of business cycles shows but slight relation 
between business activity and hog prices when activity is average 
or below; if anything, the demand is better with rather dull business 
than with moderately active business. But when business activity 
increased toward a peak, hog prices suddenly responded, an increase in 
20 See Ezekiel (5, pp. US, U6, 448) for explanation of this process. 
