台灣進口部門Armington彈性之估計與分析

Estimation and Analysis of Armington Elasticities for Taiwan's Importing Sectors

實務上常透過Armington彈性值來瞭解進口品與國產品的替代程度，而該彈性值亦是CGE模型與引力模型等模型在設定及處理跨國貿易函數時的必要參數。本文將產品生產及進口原始資料以Divisia Index合併成對應主計處產業關聯表49部門與161部門之資料，再進一步以Rolling Chow test進行參數穩定性測試，排除趨勢改變較明顯的時點，然後選取2001 ~ 2006年以及1989 ~ 1994年資料做為估計期間，再進行實證估計。在Panel Data模型估計結果中，台灣資料呈現與國外文獻類似之細分類彈性值高於粗分類，以及後期彈性值低於前期的特性；以細分類資料而言，本文所得前期平均彈性值為1.93，後期則為1.75；而以中分類而言，前期平均彈性值為1.08，後期為0.78。基於實際資料具有細分類彈性值高於粗分類的特性，學者使用CGE等模型進行模擬操作時，應考慮根據模型部門分類方式，進而選取更能反映實際經濟體系運作之參數，而彈性值隨時間遞減的趨勢，亦可做為模型動態模擬之參考。

Armington elasticity is an aggregate measure which is frequently used to reflect the substitutability between domestic and imported goods of a specific sector in an economy. It has also very often been applied to the model specification of models such as computable general equilibrium (CGE) and gravity models. This paper intends to estimate the Armington elasticities for most of the sectors corresponding to the 49- and 161-sector classifications in the input-output tables of Taiwan. We use Divisia Index to aggregate the price and quantity data obtained from the trade database, perform rolling Chow test to test the stability of parameters, exclude the years with significant structural changes, and come up with two time periods for estimation: 2001-2006 and 1989-1994. By using panel data models, our estimation results indicate that elasticity values for disaggregate sectors are generally higher than those of aggregate ones. Moreover, elasticity values show to have a declining trend as time passes. These results have important implications for modelers in determining the most suitable elasticity values for their model specifications and simulations.