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

36 The Life-Cycle Hypothesis
basis of an analysis of factors controlling these coefficients we suggest, however,
that a value in the order of .5 to .6 for b 1 and of .2 and .3 for b 2 , is not likely to
be far from the mark. Using our standard assumption as to the value of L and N,
we then get the following comparison:
Our Model
Coefficient of Klein’s estimates 64 (linear approximation)
Assets -.21; -.25 -.04 to -.03
Income (.02) (.03)
Income change .03; .07 .1 to .2
(.05) (.06)
Age .0013; .0055 .003 to .0045
(.0022) (.0024)
The age coefficient falls squarely within the range of Klein’s results and
nothing further need be said about it. The coefficient of income change estimated
from the sample is lower than we should have expected, although, at least in one
case, the difference from our estimate is well within the range of the standard
error. 65 In the case of assets, however, the statistical coefficient is clearly far too
great. A large part of this disparity is due, we suggest, to the fact that Klein’s variable
L represents liquid assets whereas our variable a represents net worth, which
is clearly, on the average, several times as large as liquid assets. Hence, if liquid
assets holdings are a reasonably good index of total net worth, Klein’s coefficient
would have to be a large multiple of ours. 66 While it is doubtful that this multiple
could be as large as 6 or 7, there seems little doubt that this correction would
cut the excess of Klein’s coefficient over the theoretical value very substantially,
probably to well within one half. 67
Unfortunately, no definite statement can be made about the coefficient of
income and the constant term on account of the logarithmic transformation introduced
by Klein. However, approximate computations we have made suggest that
these coefficients are in line with our theory in the case of his first equation and
somewhat too large (in absolute terms) for the second.
It would thus appear that, at least in terms of orders of magnitude, Klein’s findings
agree with the implications of our model. We hasten to repeat that the comparison
has very limited significance and its results must be taken with a good
deal of salt; yet Klein’s estimates, just as the many other empirical data we have
been able to locate, would seem to warrant the feeling that we are on the right
track.
Notes
1. John Maynard Keynes, The General Theory of Employment, Interest, and Money, 1936, p. 96.
2. For two excellent bibliographies in this field, see: G. H. Orcutt and A. D. Roy, “A Bibliography
of the Consumption Function,” University of Cambridge, Department of Applied Economics, mimeo-