本文建構一個縱橫分量對分量迴歸 (panel quantile on quantile regression, PQQR) 模型以檢視物流績效對出口的不對稱效果，亦即不同程度物流績效水準對不同程度出口成長的差異性效果。實證上，使用2007年至2020年期間台灣的前8大出口國之112筆縱橫資料 (panel data) 進行估計。重要的實證結果，包括：
(一)各變數對於出口成長的影響隨著不同分量的物流績效指數 ( ) 與不同分量的出口成長率 (θ) 組合而變動，並非如縱橫迴歸模型所主張的單一且固定的效果，亦非如縱橫分量迴歸模型所認定的隨不同分量的出口成長率而變動。換言之，物流績效水準高低與出口成長高低是影響各變數對台灣對這些國家出口成長的關鍵因素。
(二)極端的 ( , θ) 組合 (高或低 , 高或低θ) 能產生各變數的最大值與最小值估計係數。首先，當物流績效 (LP) 位於極高水準 ( =0.95) 且出口大幅衰退 (θ=0.05) 時，台灣主要出口國經濟成長 (GDPGR) 對台灣在這些國家出口成長 (EXGR) 的貢獻最大；當LP位於極低水準 ( =0.05) 且出口成長大幅衰退 (θ=0.05) 時，貢獻最小。其次，當 =0.55或0.60 且θ=0.05時，匯率變動率 (EGR) 對出口成長率 (EXGR) 的貢獻最大且為正值，表示新台幣貶值最有利於提升EXGR；當 =0.05或0.10且θ=0.20或0.25時，EGR對EXGR的貢獻最小且為負值，表示新台幣貶值最不利於提升EXGR。再者，當 =0.30且θ=0.90時，運輸成本提升對於EXGR貢獻最大且為正值；當 =0.05且θ=0.05~0.20時，運輸成本提升對於EXGR貢獻最小且為負值。此外，當 =0.95且θ=0.90時，改善LP對於EXGR貢獻最大且為正值；當 =0.05且θ=0.95時，改善LP對於EXGR貢獻最小且為負值。
除了改善文獻存在的缺失外，實證結果對物流成就在促進出口貿易的重要性提供一些新的認識與政策建議給出口業者、金融貸放業者、研究者，以及政府決策單位。
關鍵字：物流績效指數、引力模型、縱橫分量對分量迴歸模型、縱橫分量單根檢定
This thesis constructs a panel quantile-on-quantile regression (PQQR) model to investigate the asymmetric effect of logistics performance on exports, namely the differential effects of different quantiles of logistics performance on different quantiles of export growth. Empirically, 112 panel observations of Taiwan's top 8 export countries during the period of 2007 to 2020 are used for estimation. Important empirical results are summarized as follows:
First, the impact of each variable on export growth varies with the combination of different quantiles of logistics performance index ( ) and different quantiles of export growth rate (θ), but not the single and constant effect highlighted by panel regression model, or varying with different quantiles of export growth rate identified by panel quantile regression. In other words, the level of logistics performance and the level of export growth are key factors affecting the growth of Taiwan's exports to these countries.
Second, extreme combinations of ( , θ) (high or low , high or low θ) can produce maximum and minimum estimated coefficients for each variable. For example, when the logistics performance (LP) is at an extremely high level ( =0.95) and the export declines sharply (θ=0.05), the economic growth of Taiwan’s major export countries (GDPGR) contributes the most to Taiwan’s export growth (EXGR) to these countries, and the contribution is the smallest when =0.05 and θ=0.05. In addition, when =0.55 (or 0.60) and θ=0.05, the effect of movement in exchange rate (EGR) on EXGR is the largest and positive, and when =0.05 (or 0.10) and θ=0.20 (or 0.25), the contribution is the smallest and negative. Furthermore, when =0.30 and θ=0.90, transportation costs contribute the most to EXGR; when =0.05 and θ=0.05~0.20, the contribution is the least. In addition, when =0.95 and θ=0.90, improving LP contributes the most to EXGR and is positive; when =0.05 and θ=0.95, the contribution is minimal and negative.
In addition to improving the shortcomings in the literature, the empirical results provide some new understanding and policy suggestions for exporters, financial lenders, researchers, and government decision-making units on the importance of logistics performance in promoting exports.
Keywords：logistics performance index (LPI), gravity model, panel quantile on quantile
regression (PQQR) model, panel quantile unit root test.
Afshan, S., Sharif, A., Nassani, A.A., Abro, M.M.Q., Batool, R., and Zaman, K. (2021). The role of information and communication technology (internet penetration) on Asian stock market efficiency: Evidence from quantile-on-quantile cointegration and causality approach. International Journal of Finance & Economics, 26(2), 2307-2324.
Google Scholar
Anderson, E., and Wincoop, V. (2004). Gravity with gravitas: A solution to the border puzzle. American Economic Review, 93, 170-92.
Google Scholar
Arvis, J.F., Mustra, M., Ojala, J., Shepherd, B., and Saslavsky, D. (2010). Connecting to compete 2010: Trade logistics in the global economy. Washington, D.C., The World Bank.
Google Scholar
Ascani, A., Riccardo, R., and Iammarino, S. (2012). New economic geography and economic integration: A review. WP1/02 Search Working Paper.
Google Scholar
Ayhan, F., Kartal, M.T., Adebayo, T.S., and Kirikkaleli, D. (2022). Nexus between economic risk and political risk in the United Kingdom: Evidence from wavelet coherence and quantile-on-quantile approaches. Bulletin of Economic Research. https://doi.org/10.1111/boer.12371.
Google Scholar
