1122 The Methods of Growth Econometrics
identifying causal effects. It may seem trivial to say that the main obstacle to under-
standing growth is the small number of countries in the world, but the problem
goes beyond a fundamental lack of variation or information. It also limits the extent
to which researchers can address obvious problems, such as measurement error and
parameter heterogeneity. Sometimes the problem appears in especially stark form:
imagine trying to infer the consequences of democracy for long-run development
in poorer countries. The twentieth century provided relatively few examples of
stable, multi-party democracies among the poorer nations of the world, and so sta-
tistical evidence can make only a limited contribution to this debate, unless one is
willing to make exchangeability assumptions about nations that would seem not
to be credible.^1 As we discuss later in the chapter, the recent literature has explic-
itly sought to address this kind of problem by considering the effects of transitions
to democracy using within-country variation. This leads to some interesting find-
ings, but the short span of the available data currently precludes the long-term
assessment that will often be of most interest.
If there were a larger group of countries to work with, many of the difficulties
that face growth researchers could be addressed in ways that are now standard
in the microeconometrics literature. For example, Harberger (1987), Solow (1994)
and many others have expressed considerable skepticism about any exercise that
assumes a common linear model for a heterogeneous set of countries. In principle,
these concerns might be addressed by estimating more general models, using inter-
action terms, nonlinearities or semiparametric methods, so that the marginal effect
of a given explanatory variable can differ across countries or over time. The problem
is that these solutions will require large samples if the conclusions are to be robust.
Similarly, some methods for addressing other problems, such as measurement error,
are only useful in samples larger than those available to growth researchers. This
helps to explain the need for a flexible approach, and why growth econometrics has
evolved in such a pragmatic and eclectic fashion, drawing on a range of statistical
methods to a greater extent than is the norm in applied econometrics.
Given the small number of countries in the world, the scope for reliable evidence
is likely to rest on the use of time series variation within countries, especially as
new data become available. Many empirical growth papers are now based on the
estimation of dynamic panel data models with fixed effects, sometimes in conjunc-
tion with a time-varying “treatment” variable, such as the advent of democracy
or trade reform. The later sections of this chapter will discuss some of the rele-
vant technical issues, and the connection between some of these studies and the
microeconometric literature on treatment effects and program evaluation. This
connection not only helps to clarify the strengths and limitations of this form
of evidence, but also some of the weaknesses of the empirical growth literature
generally.
Despite the many difficulties that arise in empirical growth research, we believe
genuine progress has been made. Researchers have uncovered stylized facts that
growth theories should endeavour to explain, and developed methods to investi-
gate the links between these stylized facts and substantive economic arguments.