The second of this series of two papers is devoted to a theoretical analysi
s of spatial interaction between commodity markets. The theoretical framewo
rk that we present is referred to as the stochastic spatial arbitrage model
(SSAM); it accounts for most of the empirical regularities observed in the
first paper. Two basic mechanisms are found to be responsible for spatial
inter-market interaction, namely (i) spatial arbitrage and hedging conducte
d by traders, (ii) spatial correlation between local shocks; the latter is
favored by a similar economic and cultural environment. The SSAM includes b
oth effects and offers a wide range of predictions about price volatility,
trade, price correlations, price differentials. Statistical tests display a
convergent array of evidence in favor of the model. However several predic
tions cannot be tested by lack of statistical evidence, a circumstance whic
h shows that yet additional "experimental" work is required.