11.11.2014 Views

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

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

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

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

122 The <strong>Life</strong>-Cycle <strong>Hypothesis</strong><br />

evidence has been found that different generations <strong>of</strong> one family live on anything<br />

like the same scale (Altonji, Hayashi, and Kotlik<strong>of</strong>f [2]).<br />

All this points, to the conclusion that Barro-like behavior cannot by any means<br />

be universal, but is at best limited to a small minority. We therefore expect the<br />

coefficient l to be below unity, contrary to Barro’s view. It should instead be<br />

somewhere around Sterling’s estimate <strong>of</strong> 0.3; it could be somewhat smaller in<br />

that not all consumers are likely to be rational enough to save now in view <strong>of</strong><br />

future taxes, but it could also be somewhat higher in so far as some portion <strong>of</strong><br />

those present may be responding to the higher taxes that will fall on later generations<br />

by saving even more than called for by the Sterling estimate. Fortunately<br />

these contrasting views can be tested by relying on the estimate <strong>of</strong> the coefficient<br />

c <strong>of</strong> GS/(Y - T) in equation (4.5a).<br />

Using the definition <strong>of</strong> c we find:<br />

l =-c<br />

[ 1- f( g)<br />

]<br />

and f(g) can be approximated by the mean saving ratio (0.14 in our sample).<br />

(4.7)<br />

2 Empirical Tests for the OECD Countries<br />

2.1 Nature <strong>of</strong> Tests<br />

The battery <strong>of</strong> tests reported below covers the last three decades: the sixties, the<br />

seventies, and the eighties as far as available, namely 1987. For each country and<br />

for each variable we <strong>com</strong>pute an average value for each decade. Since there are<br />

21 countries and three decades, we have a total <strong>of</strong> 63 observations on each variable.<br />

We have performed three types <strong>of</strong> test: the first is based on a cross-section<br />

time-series <strong>of</strong> the 63 observations; the next test is based on a cross section <strong>of</strong> the<br />

21 countries for each decade separately; the third relies on a cross section <strong>of</strong> first<br />

differences. We present two versions <strong>of</strong> this test. In one the observations are the<br />

change in the value <strong>of</strong> each variable between the ’60s and ’80s; in the other the<br />

changes are between the ’70s and ’80s.<br />

2.2 Results—Level Tests<br />

Table 4.3 reports the results <strong>of</strong> a number <strong>of</strong> tests <strong>of</strong> the life cycle and <strong>of</strong> the role<br />

<strong>of</strong> fiscal variables. The first two rows test the variable that is the distinctive feature<br />

<strong>of</strong> the life cycle model, namely in<strong>com</strong>e growth g. It is seen that the coefficient <strong>of</strong><br />

this variable is highly significant, as measured by the t-ratio. In line 1 <strong>of</strong> table<br />

4.3, the estimated coefficient <strong>of</strong> g is somewhat lower than in line 2, but one may<br />

suspect some downward bias in the former estimate because in our sample we find

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!