台灣一九八零年代住宅自有率變化之探討An Analysis of Taiwan's Home Ownership Rate Changes in 1980s

本文利用民國71年及民國82年家計單位的資料(household characteristics data)米分析台灣住宅自有率變動的原因。我們首先以Probi模型來估計住宅權屬選擇模型,結果發現在該兩年當擁屋成本相對於租屋提高時,人們均領向於租屋,而預期房屋增值提高時傾向於擁屋。此外,家庭恆常所得愈高、年齡愈大、家庭人口數愈多其攤屋機率也愈高在自有率變動方面,分析11年間各解釋變數及偏好變動的影響,房價上漲帶來的成本上升與預期增值的影響大致相互抵銷。另方面房價房租比對擁屋決策影響幅度的變小包括利率下降的影響)卻是自有率提升的最重要因素。所得增加與所得對擁屋機率影窖的程度改變的影響相互抵消後是些微的負數,而人口統計變數變動的影響也大致相互抵銷。

In the 1980s, Taiwan has gone through a wave of serious house price increases.  However, the homeownership rate still moderately increased. In this research, we used household data from the "Housing Survey" of 1982 and 1993 to analyze factors that affect the home ownership rate changes in the 1980s. A probit model is used to estimate the tenure choice model for these two years respectively.  The model estimation verified our theoretical expectation that when the cost of owning a home relative to rent increases, people tend to rent and when there is an expectation of bouse price appreciation, people tend to own.  In addition, model estimation also found that permanent income increases, increases in the age of the household head and increases in household size all increase the probability of owning a home.In analyzing the home ownership rate changes in this 1 I-year span, the effect of changes in the value of variables and changes in preferences (i.e. the changes in the coefficients) are decomposed. We found that the effect of house price increases on the relative cost of owning a home and the expectation of appreciation are almost cancel each other out in terms of affecting the home ownership rate.  On the other band, the extent that relative price for owning and renting affects the ownership probability decreased during these 11 years (i.e., the absolute value of the coefficient decreased), which contributes significantly to the ownership rate increases. Income increases during this period increased the ownership rate, however, it just about to cancels the coefficient changcs. Finally, the changes in the demograpbic variables and their coefficients are also close to canceling each other out.