RELATION OF LAND [NCOME TO LAND VALUE. 18 
The average differences between the census values and sales prices 
for the counties in Figure 4 which fell within the groups listed in 
Table 2 are very small, and in no case are they large enough to affect 
the ratios of rent to value within those groups by more than 0.4 per 
cent. The greatest bias occurred in group 7, where there is good 
reason to believe that the average sales prices are less accurate than 
in other areas. In all other areas the error in the ratios due to land- 
value bias was not greater than 0.2 per cent. This bias does exist, 
however, and in particular areas is of much more importance than 
Table 2 indicates. 
There is an apparent objection to the use of average sales prices in 
1920 to check the census land values. The census values are based 
upon estimates as of January 1, 1920, which was the peak of both 
land prices and prices of farm products. Prices of farm products 
broke about the middle of the year and started on their sharp down- 
ward course. It may be supposed that land values also went down 
during the latter half of 1920, hence the yearly average sale price of 
land for this year might be expected to be below the census values, 
which represent the state of the land market at the peak of the boom. 
From data based on the monthly sales of land for 1920 in Clinton 
and Tipton Counties, Ind., however, it is evident that land values 
did not go down in the latter half of 1920, but continued upward. 
The average sale price of land in these two counties for the first six 
months and the last six months of the year are shown below : 
Per acre. 
First six 
months. 
Last six 
months. 
Clinton County 
Tipton County. 
$217 
220 
$234 
252 
It is not at all likely that the amount of increase shown on these 
two counties was general. It is more probable that land values just 
about maintained their level through 1920. It is fairly certain that 
they did not decline during that year. Land values are sensitive to 
changes in the prices of farm products, but they do not respond at 
once to upward or downward movements in these prices. 
RELATION OF CENSUS LAND VALUES TO PRODUCTIVITY. 
A further test of the accuracy of the estimated land values obtained 
by the census can be had by correlating them with productivity. If 
high land values are closely associated with high productivity and 
low land values with low productivity, then it may be concluded that 
the census land values are accurate in proportion to the closeness of 
this relationship. 
Average net cash rent is the best measure of the value productivity 
of land. Now, however, the problem is to test the accuracy o( both 
the average land values and the average cash rents. Therefore rents 
can not be correlated with values to test the accuracy of each. To 
test the accuracy of both rents and values it is necessary to correlate 
both with the factors that determine productivity. These factors 
will be referred to as productivity factors. 
