本研究運用空間縱橫模型來探討影響台灣縣市人口遷移之因素。過往的實證研究大多假設不同空間單元觀察值之彼此獨立性，空間縱橫模型則考量了人們在空間遷移時可能產生之空間相依性，也就是空間單元觀察值不再被視為彼此獨立。此外時間異質性也被時間固定效果所捕捉。實證結果顯示：平均所得水準、失業率、社會福利支出、人均居住面積能顯著的解釋縣市人口遷移行為，人口遷移之空間相依性顯著地存在於模型的誤差過程中。

This study draws upon spatial panel models to analyze factors that might influence migration across cities-counties of Taiwan. Most previous empirical studies assumed that the observations from different spatial units are independent. The spatial panel models consider the spatial dependence that may result from people moving across spatial units. In this case, observations from different spatial units are no longer assumed to be independent. Additionally, temporal heterogeneity is captured by time fixed effects of the spatial panel models. Empirical results indicate that average income level, unemployment rate, social welfare expenditure and the average area of living can significantly explain migration across cities and counties in Taiwan. The model's error process shows significant spatial dependence.