1156 The Methods of Growth Econometrics
it harder to believe some of the earlier suggestions, rarely based on evidence, that
corruption could be actively beneficial.
Given the likelihood that variables are inter-related, one response is to model as
many as possible of the variables that are endogenously determined. One promi-
nent example is Tavares and Wacziarg (2001), who estimate structural equations
for various channels through which democracy could influence development. This
approach has some important advantages in both economic and statistical terms.
It can be informative about underlying mechanisms in a way that much empiri-
cal growth research is not. From a purely statistical perspective, if the structural
equations are estimated jointly by methods such as three-stage least squares (3SLS)
or full information maximum likelihood (FIML), this is likely to bring efficiency
gains. That said, systems estimation is not necessarily the best route: it has the
important disadvantage that specification errors in one of the structural equations
could contaminate the estimates obtained from the others. Importantly, these spec-
ification errors could include invalid exclusion restrictions, a possibility that is
often hard to rule out.
24.6.2 Exclusion restrictions
The most common response to endogeneity has been the application of instru-
mental variable procedures to a single structural equation, with growth as the
dependent variable. Appendices C and D in Durlaufet al.(2005) describe a wide
range of other instrumental variables that have been proposed for the Solow vari-
ables and other growth determinants respectively, where the focus has been on
the endogeneity of particular variables. Whether these instruments are genuinely
plausible is another matter. In our view, the belief that it is easy to identify valid
instrumental variables in the growth context is often mistaken. Many applica-
tions of instrumental variable procedures in the empirical growth literature are
undermined by a failure to address properly the question of whether these instru-
ments are valid, in the sense of being uncorrelated with the error term in a growth
regression. When the instrument is invalid, instrumental variables estimates will
be inconsistent. Not enough is currently known about the consequences of “small”
departures from validity, but there are circumstances in which the 2SLS bias is worse
than the OLS bias, especially if the instruments are “weak.” It is certainly possible
to envisage circumstances under which ordinary least squares would be preferable
to instrumental variables on, say, a mean square error criterion.
A common misunderstanding, perhaps based on confusing the economic and
statistical versions of “exogeneity,” is that predetermined variables, such as geo-
graphical characteristics, are inevitably strong candidates for instruments. There
is, however, nothing in the predetermined nature of these variables to preclude
a direct effect on growth, or the possibility that they are correlated with omitted
growth determinants, and hence with the error term. Even if we take the extreme
example of geographic characteristics, there are many channels through which
these could affect growth, and therefore many ways in which they could be corre-
lated with the disturbances in a growth model. Brock and Durlauf (2001a) use this
type of reasoning to criticize the use of instrumental variables in growth economics.