Another problem is that the positive relation between investment and
growth can be strongly influenced by the presence of a few outliers. For
example a study of 29 African countries using data from 1970 to 1997 showed
that private investment had a significant, positive and strong effect on growth,
(even after attempting to control for its causality) (Devarajan et al., 2003).
However this study also found that private investment was only correlated
with growth if Botswana (being the only high-growth, high-investment coun-
try in Africa) was included in the sample. While much of this book shows how
particular case studies exhibit different patterns from the typical experience,
this is an example of how one unusual case can drive an entire statistical rela-
tion. Such problems with outliers show how the statistical findings may not
survive changes in the sample such as dropping some data points (here
Botswana) or testing different regions of the world economy or using data
drawn from different time periods. Statistical work on growth now checks the
robustness of such findings. Using a much larger sample than Africa the posi-
tive relationship between investment and growth is not found to depend on a
few outliers. Most countries in the world are high-investment/high-growth,
low-investment/low-growth or reassuringly somewhere in the middle on both
variables.
Despite these statistical problems the early development economists had hit
upon an important truth. Investment does have a real causal relation with
economic growth. In fact, no other variable has such a robust positive relation
with economic growth as does investment. Levine and Renelt (1992)
compared the impact of a number of variables that economists had suggested
were important in explaining economic growth. These included fiscal policy
indicators, international trade and price distortions, monetary policy and polit-
ical indicators. They collected data for 119 countries for the years 1960 to
- They found that only the investment share of GDP showed a consistently
positive and robust correlation with economic growth.
But still an elusive quest
Although investment has a robust and positive relation with economic growth
it does not explain everything. Other factors are more important in explaining
economic growth variations over time and between countries. For example,
The Gambia and Japan both increased their capital stock per worker by over
500 per cent between 1960 and 1985. In the Gambia output per worker rose 2
per cent and in Japan it rose by 260 per cent (Easterly, 2001b:67). Easterly and
Levine argue strongly that the attention given to investment is misplaced; it is
the productivityof the investment which matters. They say that the ‘central
problem in understanding economic development and growth is notunder-
standing the process by which an economy raises its savings rate and increases
the rate of physical capital accumulation’(2001:177). Easterly (2001b) does
not accept that the link between investment and growth is robust or strong
enough,and argues that there is still an ‘elusive quest for growth’.
66 Sources of Growth in the Modern World Economy since 1950