and the hypothesis that λ equals 0.3, roughly the average found in other consumption studies,
could not be rejected. Therefore it seemed appropriate to assume the same share of liquidity
constrained consumption of 0.3 in all countries. The pure rate of time preference θ was
determined from the relationship between wealth and consumption that is implied by the life
cycle model1. As can be seen in Table 1, it implies a relatively low rate of time preference in
high saving countries like Japan and a larger rate in low saving countries like the US.
Table 1: Parameters of the consumption function
country |
probability of p |
rate of time |
share of liquidity |
~BL |
002 |
0.0092 |
03 |
DK |
0.02 |
0.0076 |
0.3 |
DE |
0.02 |
0.0088 |
0.3 |
GR |
0.02 |
0.0096 |
0.3 |
ES |
0.02 |
0.0092 |
0.3 |
FR |
0.02 |
0.0092 |
0.3 |
IR |
0.02 |
0.0092 |
0.3 |
IT |
0.02 |
0.0092 |
0.3 |
NL |
0.02 |
0.0084 |
0.3 |
OS |
0.02 |
0.0084 |
0.3 |
PO |
0.02 |
0.0092 |
0.3 |
SF |
0.02 |
0.0084 |
0.3 |
SW |
0.02 |
0.0078 |
0.3 |
UK |
0.02 |
0.0080 |
0.3 |
US |
0.02 |
0.0100 |
0.3 |
JA__________ |
__________0.02__________ |
__________0.0052__________ |
_______________0.3_______________ |
Notes: * implies a forward looking horizon of 50 years (1/p)
** same coefficient imposed for all countries. Individual country estimates ranged from 0.2 to 0.4 on
quarterly data and 0.2 to 0.6 on annual data (estimation period: 1975-1996)
1 The path of consumption in such a model can be written as C= (∣- - (θ + π))<' - (θ + p)nFW where π
is productivity growth (see eg Buiter (1988)). It follows that along the steady state growth path
θ = [(r - n)& - sq):]/(& + q):). The value of θ in Table 1 shows the mean value of this
expression for each country over the 1990s, with the exception of the cohesion countries for which a
small adjustment has been made to capture the relative state of their convergence.