INPUT AS RELATED TO OUTPUT 39 
Other parts of the plan, such as the number of cows to b^ kept, 
the acres of corn to grow, can be calculated at much shorter range. 
Plans based on input variations and fitted to the particular condi- 
tions of the farm should give a much closer forecast of actual re- 
sults than plans based on inflexible requirements. 
COST INDICES 
For certain purposes it is desirable to have an index figure show- 
ing how the costs of various farm products have changed in the 
past, what they are now relatively, and, in some cases, what they are 
likely to be in the near future. The plan usually suggested for 
making such indices is to apply cost rates at the time to unit re- 
quirements of the various cost factors, and thus to assemble a series 
of total cost figures which can, if desired, be reduced to relatives 
on the basis of some one year's costs. 
It must be realized that the purpose of such indices is not to show 
what costs should have been, but rather what they were and are. 
At any one time most farmers are producing at other than the least- 
cost or most profitable combination of inputs; also the limitations 
prevailing on each farm make the least-cost and most profitable 
combination different for different farms. For an index of past 
changes, therefore, some summary expression of actual inputs is 
what is wanted. 
If the index, however, is to be in any way balanced against price 
indices for the same products, it should not be based upon the modal 
or mean inputs, but upon the inputs of the marginal group of pro- 
ducers, say, all those whose product falls between the upper sixty- 
fifth and ninety-fifth percentiles. This will give a group large 
enough for the necessary stability of the index, and yet reasonably 
responsive to changes in conditions of production among the mar- 
ginal producers. The producers between these two percentiles will, 
in some cases, be the poor farmers producing all products at high 
costs; but they will mostly be better farmers growing so large an 
acreage of this particular crop that their alternative costs are high. 
The upper 5 per cent, it might be 10 per cent, are excluded because 
they probably represent abnormal conditions, such as accidentally 
low yields and the like. 
If indices are to be used in forecasting necessary price, the index 
of marginal inputs is the better one to use, although it would be 
desirable for such a purpose if the inputs of all-round poor farmers 
with no better alternatives could be kept out of the group average. 
For some purposes, cost indices based on least-cost inputs would 
be more useful. 
The principal difficulty in constructing cost indices is that of cost 
rates. If flat rates are applied to all labor, for example, the labor 
element will be overweighted or underweighted in the index. Charg- 
ing family labor at hired labor rates has the same effect. Similarly 
the land may be overweighted or underweighted by flat rent charges 
for all crops. 
Will such indices be of use to farmers in choosing combinations of 
enterprises? If a series of cost indices were projected into the 
coming year on the basis of probable cost rates, farmers could note 
any trends, compare them with indices of* trends in prices of the 
