328

BANKING STANDARDS
in determining the level of net earnings at a given time, to consider
 the relative levels of both gross earnings and of total expense
 ratios, the relation between which determine net earnings.
But the level of the ratios in both series in a given year are
functions of their levels in the preceding year. This fact has been
demonstrated above, not only for the ratios for individual banks
in a given district but also for the total member banks by districts.2®
 Accordingly, it must be taken into account in determining
 the relative effect of variable gross earnings and of variable
total expense ratios on ratios of net earnings.
Data showing ratios of gross earnings, of total expense, and
of net earnings are available for 1924 and 1925. Let us trace
out the regression pattern, as it relates to each of these types of
ratios, in terms both of levels and of year-to-year changes, and
pbserve the results.
Between 1924 and 19235, gross earnings, total expense, and net
earnings, as percentages of earning assets, regressed to type, the
amounts of change varying directly and the nature of the change
inversely with the amounts and the positions of the ratios relative
 to their 1924 average. Stated in another form, and without
respect to averages as points of reference in the three series,
ratios which were low tended to increase and those which were
high tended to decrease between 1924 and 1925, the amounts of
change being functions of the size of the ratios themselves in
1924. The net earnings ratios in 1925, therefore, are functionally
related to these regression tendencies. The same condition, it
is believed,2® obtained in 1924, although data for establishing the
fact are not available.
If, as between 1924 and 1925, ratios of gross earnings, which
are high in 1924, tend to fall, the net earnings of the banks in
question, other things being equal, will tend to fall. Contrariwise,
if those which in 1924 are low tend to rise, the net earnings of
the banks in question, other things being equal, will tend to rise.
But the other things are not equal, inasmuch as total expense

25 Gee the discussion of regression to type in chapters relating to these ratios,
passim.

26 Confidence in this belief is found in the similarity of regression tendencies for
banks within and between districts. As between districts for the three series, rezressions
 of the nature indicated characterized the years 1919 to 1925; as within
districts, they obtained for the item of expense in the Boston district between 1922
and 102%, and in the New Vork district between 1923 and 1925.