it would cause the supply to
wither up and disappear.

Sec.4. _Risk under Large-scale Organization_. While this is true of the
present economic system, the question is worth considering whether it
represents a fundamental necessity, whether, for instance, under our
world socialist commonwealth the factor of risk-bearing need play so
important a part as it does in the actual business world. This
question cannot be answered with a conclusive simplicity; opposing
considerations present themselves, between which it is not easy to
strike a balance. On the one hand, in accordance with the law of
averages gains and losses tend to cancel out over a large series of
transactions, _when reasonable calculations have been made_. Thus
Insurance Companies, while they take heavy risks off the shoulders of
policy-holders, incur relatively trifling risks themselves; they can
predict the aggregate sums which they will be called upon to pay
within a small margin of error. In the same way it might seem that
every enlargement of the scale of business would make for an automatic
insurance and a consequent economy of risk; and thus that if all
businesses were comprised in a single financial unit, gains and losses
would cancel out over so wide a range that the degree of risk
remaining would be almost negligible.

This might indeed happen, if business risks were mainly of that
objective kind in which the insurance companies specialize; for then
we could assume that the chances of success or failure would be
estimated reasonably. But, in fact, most business risks, not being of
this kind, must be estimated by processes of human judgment, which are
very fallible. And here we must take account of the law of averages in
another aspect, with a different bearing on the argument. When an
industry comprises a large number of separate concerns, and the
decisions accordingly are taken by many men, acting independently of
one another, the errors of calculation will tend to some extent to
cancel one another out. The undue optimism of one man will be balanced
by the undue pessimism of another; and, if there is no prevailing bias
in either direction, the errors of judgment will not affect the
results for the industry as a whole. But where the effective decisions
are taken by very few men, the chances are far greater of a
preponderating balance of error in one direction. The risks dependent
on the factor of human judgment tend therefore to increase.

This