50 BULLETIN 1440, U. S. DEPARTMENT OF AGRICULTURE 
price of hogs are graphed. Finally, the effect of the supply of hogs 
upon the price received is represented in the four regression curves, 
contributing some part to the shape of each. 
The curves are of such shape that three or more constants would 
be required to state each as an analytical function. That means that 
the error in determining each such constant by multiple correlation 
would be higher than simply using four variables would indicate. It 
may be that some of these peculiarities of shape represent merely 
particular relationships between the variables during the rather short 
series of years studied, which had no real causal significance, and 
which therefore will not be repeated hereafter. Although this is 
admittedly a possibility, there is no way of testing it out in advance ; 
all that can be said is : " These relations did exist in the past. Barring 
other forces coming into play, it seems reasonable to expect them to 
be likely to reoccur in the future." 
No attempt is made at this point to give a logical explanation of 
the significance of each of these curves. The whole body of the 
previous analysis has developed the nature of the relationships 
involved; the attempt here is merely to obtain empirically a method 
of price forecasting, without attempting specific explanations of the 
functions obtained. 
The application of these results to forecasting hog prices under 
present conditions is limited by three unknown factors. 
(1) What is the postwar trend of production? 
(2) What is the postwar trend of hog prices? 
(3) To what extent will the pig-survey information change the 
market's reactions? 
As has already been developed, the postwar trend of demand has 
not yet become sufficiently stable to make possible an estimate of 
its future course. Barring that, question 2 must remain unanswered 
for the present, and until that question can be answered, question 3 
can not be investigated with any degree of accuracy. 
FORECASTING THE PRICE OF HOGS BY PERCENTAGE CHANGES 
An attempt was made to obtain a "forecasting formula" to fit 
postwar conditions by using percentage first differences instead of 
deviations from trend, and so remove the problem of specifically 
determining trends. Obviously, however, the differences themselves 
included whatever there was of a regular trend effect. 
The following variables were used: 
F Corn-hog differential-?- Bureau of Labor Index, lagged 18 months. 
G Corn-hog differential-f- Bureau of Labor Index, lagged 6 months. 
H Index of Prices of Industrial Stocks, lagged 6 months. 
Price of hogs, corrected for normal seasonal variation, then divided by Bureau 
of Labor Index 
Moving average of same figure for 18 months previous 
The price of hogs, I, was thus stated as the per cent which the 
price, adjusted for price level and seasonal variation, was of the 
moving average of the same price 18 months previously. In other 
words, the dependent variable was the percentage change in price 
over an 18-month period. 
For the period, June, 1905, to May, 1913, the correlation between 
the corn-hog differential and this per cent change, r FI , was equal to 
+ 0.85, and the multiple correlation of the three variables with the 
