The implicit theorizing in all of these models has obvious shortcomings in explaining fluctuations in the level of
economic activity in industrialized countries. But these are even less satisfactory as macroeconomics in the
context of developing countries, where agriculture creates a dualism, where the financial sector is
underdeveloped, where the informal sector is large, where markets are not perfect, where prices are often
flexible and where unemployment, often disguised as underemployment, is widespread.2
The macroeconomic aggregates are, of course, the same. So are the macroeconomic identities. For an
understanding of macroeconomic systems, however, the accounting relations of aggregates need to be
combined with an economic analysis of causal determinants, to describe the behaviour of households, firms and
governments. It is here that differences arise. The nature of relationships (between variables) and the direction
of causation (what determines what) are both a function of the setting or the context.
The starting point for any macroeconomic analysis is the distinction between exogenous and endogenous
variables or that between autonomous and induced changes. Such a distinction is essential in macroeconomic
theorizing which seeks to analyse policy implications. It is important to recognize that this distinction is derived
not from the analytical structure but from the institutional setting of models.
The most important example, perhaps, is the Keynesian idea that investment is an independent (exogenous)
variable to which saving adjusts as a dependent (endogenous) variable. Investment is autonomous, determined
by profit expectations of firms, while saving is induced, determined by income of households. The distinction
rests on the institutional assumption that firms have access to credit from commercial banks and financial
institutions depending on their credit worthiness and the expected profitability of their projects, irrespective of
the level of savings by households in the economy. Thus, it is the institutional setting of what Hicks
characterized as an 'overdraft economy' which allows investment to be financed in advance of, and independent
of, the level of saving in the economy.3
A short digression is worthwhile. If credit is endogenously determined by demand, then the Keynesian
perspective emerges in its sharpest focus. The independence of the investment function of firms determines
effective demand (hence output), while the financial system merely plays an accommodating role without
influencing the level of demand (hence output). On the other hand, if credit is assumed to be exogenously
determined by the financial system, the monetarist perspective, or at least a perspective that seeks to focus
more on finance, emerges as critical, in so far as banks have a role in influencing the investment decisions of
firms by relaxing or tightening credit facilities extended to firms.4 Or, if firms have to rely mostly on their own
capital to finance investment in a financial market without depth, then corporate profits, or savings, largely
determine investment.5 In either case, investment can no longer be treated as exogenous.
The moral of the story is clear. The distinction between exogenous and endogenous variables, or that between
the autonomous and induced changes, is essential in macroeconomic models, which seek to analyse events or
prescribe policies. However this distinction does not derive from the analytical structure of a model but from the
17 April 2015 Page 2 of 13 ProQuest
underlying institutional setting of the model. Much can change, especially in terms of policy prescriptions, when
institutional settings, hence determinants of causation, change. And without some notion of causation, or the
sequence of events in time, no policy prescription is possible. Therefore, even equilibrium relations require
some causal interpretation, at least for policy analysis.
It follows that macroeconomics developed in the context of industrialized countries cannot simply be
transplanted in developed countries. Starting from accounting identities, models can be built based on economic
reasoning, but such models must respect institutional facts and recognize different contexts.