Interregional Migration Modeling via Entropy and Information Theory

Abstract

INTRODUCTIONAs demonstrated by Rogers (1966) and further analyzed by Mueser and White (1989), the steady-state growth solution of a Markov-type migration model with transition probabilities constant over time yields, for each region, the in-migration rate equal to the out-migration rate (plus a small local growth correction). Net migration models, which have dominated the literature because of the concern of policymakers with net population changes in estimating infrastructure requirements, become of less interest as gross in-flows and out-flows become more nearly equal. Also, the "net migrant" is not behaviorally identifiable (Rogers 1989). The models of gross migration, from which net migration totals can be extracted, are based on the techniques of entropy or information theory, such as in the work of Plane (1982), Rogerson and Plane (1984), Plane and Rogerson (1985), and Holm-

Entropy‐Based Spatial Interaction Models for Trip Distribution. 基于熵的空间相互作用模型在出行分布中的应用

“… 5 Among the many papers in Geographical Analysis reviewing or citing Wilson's work are: Batty (1976a), Berry and Schwind (1969), Cadwallader (1987), Chan (1992), Coelho et al (1978), Coelho and Wilson (1977), Diplock and Openshaw (1996), Fotheringham and Knudsen (1986), Fotheringham and Webber (1980), Haynes and Phillips (1987), Jefferson (1977), Johnston (1982), Lo (1991), Marchand (1974a), Marchand (1974b), Morrow‐Jones (1980), Rho et al (1989), Roy and Flood (1992), White (1982), and Wilson and Birkin (1987). …”

Gravity and Spatial Structure: The Case of Interstate Migration in Mexico*

“…The GME estimator is a member of the class of information‐theoretic estimators, which is based on the entropy‐information measure introduced by Shannon (1948). The use of entropy‐based methods has already a long history in the migration literature, ever since the seminal work of Wilson (1970) and subsequent papers by Plane (1982), Roy and Flood (1992), and others. However, in this paper the entropy approach is applied within a regression framework, which is, therefore, fundamentally different from conventional Maximum Entropy (ME) applications in earlier studies of migration—which are primarily concerned with “matrix balancing” (i.e., recovering, predicting, or updating matrices of bilateral migration flows) based on incomplete information.…”

Entropy in Urban and Regional Modelling: Retrospect and Prospect. 城市和区域建模中的熵：回顾与展望

“…Models of trade flows are also important and more interesting, because in this case, there are additional constraint equations to those that determine the flows—the input–output relationships. The model‐building principles still apply, and an integrated model can be built (Wilson 1970b and extensively developed, particularly by Hewings; see, e.g., Kim, Boyce, and Hewings 1983; Roy and Flood 1992; Oosterhaven 2005).…”