Regression Analyses of Migration with an Application to Selected Provinces in Thailand

Abstract

Mathematical models of migration are developed that are suitable for describing migration in Thailand of people in four different occupation classes. Nine main types of human migration models and the data that they require are reviewed. However, data is available from the National Statistical Office of Thailand for only three of these models: Gravity, Expanded Gravity, and Neo-Classical/Labor Flow Models. The available Thai data does not fit any of these models in their original form. A linear regression analysis on logarithmically transformed data is carried out to find the variables that are most important for describing migration for provinces from four main regions: Central (Bangkok), North (Chiang Mai), North-East (Nakhon Ratchasima) and South (Songkhla). We analyze movement out from, and movement into, these four provinces for all other provinces for which data is available. The most important variables for moving out of Bangkok are unemployment (or labor force) in the destination province and distance to that province. For moving into Bangkok only labor force in the province of origin is important. For Chiang Mai, wage rate is the most important with distance being less important for moving out and cost of living or wage rate are the most important for moving in. For Nakhon Ratchasima, wage rate and labor force are important for moving out and labor force and distance for moving in. For Songkhla, labor force or wage rate are the most important for moving out. The results for moving in to Songkhla are not reliable. The results show that the reasons for migration for the four occupation classes can be different and that the reasons for moving in and moving out of each of the four provinces can also be different.