中國城鎮食品需求之估計：追蹤資料模型之運用

Estimating Food Demand in Urban China: A Panel Data Approach

本研究旨在運用追蹤資料模型來估計並探討中國大陸城鎮地區之食品需求。在使用1992–2001 年29 個城鎮地區之省級資料以及追蹤資料模型中的固定效果模型（包含地域、時間以及兩者兼具的效果），本文估計了九種主要食品（包括糧食、油脂、豬肉、家禽、水產品、蛋、奶及奶製品、蔬菜與水果）的近似理想需求體系。實證結果顯示：追蹤資料模型比混和資料之模型更能解釋中國大陸城鎮地區龐雜分歧的食品需求。本研究進一步發現：這九種食品皆符合需求法則並對價格變動較不具彈性。其次，除了蛋在廣東及海南為劣等財（所得彈性為負值）之外，其餘皆為正常財；而部分的水產品甚至為奢侈品（所得彈性大於1）；再者，省市間差異亦顯現在所得彈性上。因此，中國大陸城鎮地區歧異的食品需求，將引伸出不同的農業發展策略與政策意涵。

This empirical study estimates the AIDS model by employingChinese urban food consumption data at the provincial level during theperiod 1992 to 2001. Nine food groups are selected in this study,including grain, oil, pork, poultry, aquatic products, eggs, milk and dairyproducts, vegetables and fruits. The methodological approach followedin this study is the fixed effect model of panel data, which were dividedinto three parts: regional dummy, time dummy, and both combined. Theempirical results show that the panel data approach performs better thanthe pooled data model; moreover, some intriguing results are found inthis study. First, the own-price elasticities of all food are between 0and –1, indicating that all food groups are not only consistent with the lawof demand but also less elastic as food price changes. Second,according to the income elasticities, most of the food groups are normalexcept eggs in Guangdong and Hainan in 2001 with negative incomeelasticities. Third, regional differences are also shown in the variousincome elasticities. Therefore, the heterogeneous food demand inurban China may stimulate several strategies in agricultural developmentand policy implications.