This paper advances a line of work on an expected-value model of socia
l exchange, in which a power structure indicates opportunities for exc
hange and a sample space of exchange networks. When the probability di
stribution of the networks in this sample space is known, the expected
-value model provides an excellent account of the distribution of bene
fits among actors in a variety of power structures. The model would be
more elegant if it also predicted the probability distribution of exc
hange networks. Furthermore, in its current form, the model fails to a
ccount for the observed exchange payoffs in the Kite, a structure that
has attracted considerable attention among exchange theorists. Here I
advance the expected-value model by deriving the probability distribu
tion of exchange networks from a simple process in which the probabili
ty of an exchange at time t depends on the value of an exchange at tim
e t-I. I show that this approach addresses the anomalies posed by the
Kite structure.