| 摘要: | 台灣地區行動用戶之普及率在2022年高達130%，意味著平均每個人平均擁有超過一個手機門號，從2004年到2022年，台灣行動電話用戶數自22百萬戶成長至30百萬戶，成長比率高達36%，在這段時間裡，台灣手機產業經歷了巨大的變革與發展，手機不僅僅是通訊工具，更成為了人們生活中不可或缺的一部分。
本研究旨在深入探討近二十年來台灣地區家用手機需求與家庭通訊費之分析，首先使用Poisson模型探討家計本業薪資對家用手機需求影響，並討論影響家用手機擁有的決定因素，其中解釋變數包含家庭所得、所得人口數、65歲以上人口數及成年人口數；考量手機數量的內生性問題，本研究除以一般迴歸(OLS)估計手機擁有數量對家庭通訊費支出的影響，並採用二階段最小平方法(2SLS)修正內生性。
實證結果顯示：家庭所得、所得人口數及成年人口數對家用手機擁有數呈正向關係，惟65歲以上人口數呈負向成長且逐年遞減，經Chow Test統計檢定不存在年度間結構性轉變之情形；另從家計通訊費與家用手機內生性及外生性實證得知，家用手機擁有數對家計通訊費則有正向之影響。另外手機內生性以所得為工具變數所得出的係數較手機外生性為大，經以Hausman test檢定結果除2021年不存在內生性外，其各年度存在內生性，故以二階段最小平方法為較適模型；手機內生性以所得及人口特性為工具變數所得出的係數與外生性相當，但Hausman test檢定結果內生性與不存在內生性的年度各半，無法說明二階段最小平方法為最佳估計。相較於放入所得及人口特性為工具變數之二階最小平方法，所得是最重要的變數，影響手機擁有數量多寡，主要來自於所得的高低，將所得單獨作為工具變數在估計上可能更具有效性。;In 2022, Taiwan′s mobile user penetration rate soared to 130%, indicating that, on average, each individual possesses more than one mobile phone number. From 2004 to 2022, the number of mobile phone users in Taiwan surged from 22 million to 30 million, representing an impressive 36% increase. This remarkable growth underscores the dynamic evolution of Taiwan′s mobile phone industry. Mobile phones have evolved beyond mere communication devices to become indispensable components of people’s lives.
This study aims to conduct an in-depth analysis of household mobile phone demand and household communication expenses in Taiwan over the past two decades. To achieve this goal, the research methodology first utilizes the Poisson model to explore the impact of household primary income on household mobile phone demand and discusses the factors influencing household mobile phone behavior. The explanatory variables include household primary income, income-earning population, population over 65 years old, and adult population. Then, an ordinary least squares (OLS) regression analysis is conducted to investigate the impact of household communication expenses on household mobile phone demand, considering the endogenous issue of the own number of mobile phones. Therefore, a two-stage least squares method (2SLS) is adopted to estimate and correct for endogeneity.
Based on the empirical results, household income, the number of income earners, and the adult population exhibit a positive relationship with the number of household mobile phone ownership. However, the population aged 65 and above shows a declining trend year by year. Besides, the Chow Test statistics indicate no evidence of structural changes.
Furthermore, from the empirical evidence of endogeneity and exogeneity analysis of household communication expenses and mobile phone ownership, it is found that the number of household mobile phone ownership has a positive impact on household communication expenses. Additionally, the endogeneity of mobile phones, using income as an instrumental variable, yields a coefficient larger than that of exogeneity, except for the year 2021. Thus, the two-stage least squares method is deemed more appropriate.
Regarding the endogeneity of mobile phones using income and demographic characteristics as instrumental variables, the coefficients are comparable to those of exogeneity. However, the results of the Hausman test indicate that endogeneity exists in half of the years, which does not conclusively determine the two-stage least squares method as the best estimates. Compared to the two-stage least squares method with income and demographic characteristics as instrumental variables, income appears to be the most significant variable affecting the quantity of mobile phone ownership. Therefore, using income as a standalone instrumental variable may be more effective in the estimation. |
