The Markov analysis of residential mobility patterns in low-cost housing in the Bangkok metropolitan areas : case study in the area of Bangkhen-Don Muang, Bang Kapi, Phra Khanong and Nonthaburi

Abstract

The first part of the study is to apply the Markov technique to analyze the residential mobility patterns only in a case study of four types of low-cost housing. Those are row house/townhouse, condominium, up-graded slum housing and housing in non-graded slum. All residences are located in the districts of Bangkhen-Don Muang, Bang Kapi, Phra Khanong and Nonthaburi. The total sample is 1,400 households, and the informants may be chiefs of households or spouses. The residential mobility patterns here are those which occur on average 10 years per time.

The one step transition probability of residential mobility was derived from the numbers of samples who are now staying in one of the four types of low-cost housing and had their former residence in one of th four types of low-cost housing.

The logistic regression was employed to analyzed the factors affecting residential mobility. The difference of salary is a factor which appears to influence the movers in almost every group, which indicates that money is the most important factor influencing low income group mobility.

From logistic regression, the simulation technique was used to determine whether a person is a mover or nonmover among four types of low-cost housing. The probability of residential mobility was then developed.

When comparing the probability of residential mobility analyzed by a logistic regression model and the Markov model, it is found that the values have rather difference. This phenomenon may be explained by the factors which are used to analyze with logit. They are all expected variables which are estimated to be influenced to the individual's decision whenever he changes his dwelling but the true decision to move or not move may come from other varialbles which are not taken into the model. The decision process is rather complex. It depends on the individual who has to make the decision and also depends on situations, environment and time. However, to use Markov to calculate the probability of residential mobility will not isolate the factors affecting the movers but will analyze instead the result of the individual's decision.