Answer:
One of the basic arguments linking political centralization with economic
reward rests upon the desire of people to benefit from the gains in welfare
which can be reaped from markets. In essence, the argument is Ricardian... the contribution of the state is to provide order and peace and
thereby to render production and exchange possible for members of society. The origins of the state, then, lie in the welfare gains that can be
reaped through the promotion of markets.
He suggests that gains from trade are greatest where products from one ecological zone
can be traded for products from another. It is near ecological divides, then, that we
should expect to see states. To support his view, he takes 34 pre-colonial African societies, asks whether they “abut an ecological divide,” and classifies them as having a
“kinship” political structure, “chiefs,” or “central monarchs.” I present a condensed version of his results in Table 1. The proportion of societies with central monarchs is greater
on an ecological divide.
In this paper, I argue that Bates (1983) is ultimately correct. His argument has been
overlooked because his sample size prevents him making a credible econometric argument that this correlation is causal. In this paper, I use ethnographic and geographic
data to overcome this limitation. I take data on state centralization for 440 societies
in pre-colonial sub-Saharan Africa from Murdock’s (1967) Ethnographic Atlas. Merging
the map of African ethnic groups from Murdock (1959) with information on African ecological zones from White (1983), I am able to compute for each society an index of its
“ecological diversity,” which I take as a proxy for the gains from trade that existed before
colonial rule. I show that this index is strongly related to the presence of pre-colonial
states. I use spatial variation in rainfall to control for possible reverse causation, and
show that the OLS estimates of the impact of ecological diversity are not overstated. I
also use exogenous geographic features to predict raster-level ecological regions, and
find that the diversity measured by these predicted points is also related to pre-colonial
African states. The relationship between trade and states is robust to several additional
controls, removing influential observations, alternative measures of states and trade,
and a variety of estimation strategies.
I show that the “Ricardian” view better explains the relationship between states and
diverse ecology than six alternative stories. First, while larger territories may require
more levels of administration and may be more diverse, area does not explain away the
relationship between ecological diversity and states. Second, because panel data are not
Explanation:
