"Life Cycle" Hypothesis of Saving: Aggregate ... - Arabictrader.com

Utility Analysis and the Consumption Function 33
bear any definite and stable relation to one another. We have further shown elsewhere
61 that the individual marginal propensity to save bears, in turn, a very
complex relation to the time-series marginal propensity and hardly any relation
at all to the time-series average propensity.
Needless to say, many implications of our theory remain to be tested. In fact,
it is a merit of our hypothesis that it leads to many deductions that are subject to
empirical tests and therefore to contradiction. We would be the first to be surprised
if all the implications of the theory turned out to be supported by future
tests. We are confident, however, that a sufficient number will find confirmation
to show that we have succeeded in isolating a major determinant of a very
complex phenomenon.
Appendix: Some Suggestions for the Adaptation of our Model to
Quantitative Tests and Some Further Empirical Evidence
Of the many cross-section studies of saving behavior in recent years, the one that
comes closest to testing the hypothesis represented by our equations (II.1) and
(II.2) is the extensive quantitative analysis reported by Lawrence R. Klein in
“Assets, Debts and Economic Behavior,” op. cit. (see especially pp. 220–227 and
the brief but illuminating comment of A. Hart, ibid., p. 228). In this appendix we
shall attempt to provide a brief systematic comparison of his quantitative findings
with the quantitative implications of our theory.
We have already pointed out the many shortcomings of his analysis for the
purpose of testing our theory, both in terms of definitions of variables and of the
form of the equation finally tested. Because of these shortcomings, the result of
our comparison can have at best only a symptomatic value. In fact, the major
justification for what follows is its possible usefulness in indicating the type
of adaptations that might be required for the purpose of carrying out statistical
tests of our model.
Making use of the identity
N -t
∫
NL ( - t) Mt - ,
L L
our equation (II.2) can be altered to the form
s
i
L-
ti
NL ( - ti)
= yi
- y
L LL
i
i
i e
ai
M
- +
L LL ty i i e ,
i
i
(A.1)
where the subscript i denotes the i-th household, and L i is an abbreviation for
L + 1 - t i .