Limitations of the Mathematical Method 105 



understanding of the normal working of supply and demand to use 

 both hyperbolas and cubic curves on occasion. 



Other refinements of the mathematical study of hog prices may 

 consist in working out the correlation coefficients between hog 

 prices and receipts at six markets or eleven markets instead of 

 using Chicago receipts alone. Work may be done looking into 

 the relation between hog prices and potential supply as contrasted 

 with the temporary or month-by-month supply. So far as the 

 relation between hog prices and business conditions is concerned, 

 it should be worth while to work out correlation coefficients be- 

 tween hog prices and the amount of new building, or hog prices and 

 Dun's index number. In fact, there are a great many measures 

 of business activities which may possibly measure the demand for 

 hogs bettor than bank clearings outside of New York City.* 



Some people ma}'^ think it advisable to work out a correlation 

 and line of regression illustrating the relation between hog prices 

 and corn prices. This has been attempted, but it has been found 

 that after the secular and seasonal trends are taken out of both 

 corn prices and hog prices there is practically no relation between 

 them. It is a curious commentary on our present marketing sys- 

 tems that corn prices and hog prices, wliile very closely related 

 decade by decade, liave ver}^ little influence on each other month 

 by montli. In other words, changing costs of production can have 

 practically nothing to do with the month-by-month changes in the 

 market price under our present economic system. Unusually high 

 com prices today are more likely to influence the hog prices of next 

 3'ear than the hog prices of today. 



After everything has been done which can be done by mathe- 

 matical method, there will still be room for common-sense judg- 

 ment. But such judgment is best applied by men wise in market 

 lore, men familiar with the technique of production, and who also 

 are familiar with such mathematical methods as are here described. 



Since the chapter, "Limitations of the Mathematical Method," was 

 written, it has heen discovered that hog receipts at eleven markets are a 

 more accurate indicator of hog prices than receipts at Chicago, and that 

 prices of Connelsville coke are a better indicator of the demand for hogs 

 than bank clearings outside New York City. The multiple coefficient of 

 correlation between hog prices on the one hand and Chicago hog receipts 

 and bank clearings outside New York City on the other hand is .65, whereas 

 between hog prices and hog receipts at eleven markets and coke prices 

 the multiple coefficient of correlation is .70. 



