- 37 -


probabilities from the new technology on the other. Critically important in the

latter is the fact that the probabilities are those provided externally to his

experience -- the extension agent tells the farmer that the average yield from

the new practices will be "x," but the farmer does not necessarily know that this

is so from his own directly observable experience. In the Knight sense, what may

appear to the extension agent as a risk in the calculable probability sense based

upon experiment station data, becomesfor the peasant farmer an "uncertainty."

Utilizing our earlier terminology, both the subjective perceptions and objective
 of the farmer and the extension agent
realities/diverge. A critical element thus becomes the "degrees of belief" or the

firmness of reliance which the farmer can place upon the estimated probabilities

associated with any particular decision for a new course of action.

 Let us assume that we are dealing with a type A subsistence farm family which

consumes 80 per cent or more of their annual average farm production. Let us

further oversimplify by assuming that this 80 per cent of production equals the
 1/
total bundle of consumption goods which constitutes the level of living for the

family. In good years and in bad the amount of production reserved out of the

staple food production on the family farm remains fairly constant. Thus, in good

years he sells a higher fraction of his total production and in poor years, less.

The critical factor is that at all times he seeks to preserve or guarantee the

minimal subsistence living of his family out of production; consequently, the

absolute minimal production reserved for family consumption remains fairly con-

stant. This level of consumption is related to my concept of a minimum subsistence


1/ We could use a food staple as rice as the sole product or numeraire.

2/ This statement is not quite accurate as is shown in my earlier formulation
where the current minimum subsistence standard is a function of a number of vari-
ables including previous levels of income. It also oversimplifies in.ignoring the
obvious effects of the income elasticity of demand for farm produced food which
though declining as development proceeds is still quite high for low income
producers [Stevens, 1965, p. 55].