以縱橫資料及背包問題求解技術探討國家基礎建設最佳投資分配(II)

Abstract

我國基礎建設推動，係由主辦機關提出先期規劃構想(或可行性研究)及總工程建 造經費概估函報主管機關審查核准，此審查程序未能與國家競爭力連結。過去基礎建 設投資及分配策略之相關研究文獻，常僅對公共投資之總投資金額及與GDP 之關係 進行探討，並未再深入探討各項建設（如鐵路、道路）之投資分配。部份探討分配策 略之相關研究亦以專家訪談及問卷為主，缺乏定量數據之客觀分析。 本研究第一、二年從提升國家競爭力之觀點，利用縱橫資料分析技術(panel data)，探討與我國國家特徵相似、且其GDP 表現及競爭力較高之先進國家其基礎建 設差異，決定我國基礎建設之重點投資項目，再利用背包問題(knapsack problem)之求 解分析技術（固定效果分析）及遺傳演算法，探討如何使有限之政府預算達到最有效 率之投資分配，建立符合國家競爭力所需之基礎建設投資及分配方式。第三年並利用 1994 至2008 年政府實際基礎建設投資金額及瑞士洛桑管理學院（IMD）國家競爭力 排名，驗證本模式之配適度及預測力，最後建立國家基礎建設最佳投資分配系統，提 供政府規劃及發展國家基礎建設計畫之參考。

In Taiwan, the infrastructure development plan is an integrated effort of several responsible bureaus and is sent to the administrative government for review and final approval. The initiation and review process does not take national competitiveness into account. Most literature on infrastructure investment focused on the total infrastructure expenditure and its relationship with national GDP, and did not discuss the distribution of each major infrastructure item (e.g., railroad, road). Some papers addressing infrastructure investment distribution problem drew their conclusions mostly based on experts』 opinions using interview or survey techniques. Such conclusions lack quantitative and objective analyses. This research attempts to establish infrastructure investment distribution model from the perspective of national infrastructure competitiveness. In the first and second year, a panel data analytical technique will be used to compare the infrastructure expenditures of Taiwan and other similar-size countries that performs competitively in terms of GDP and IMD competitiveness score to determine the key infrastructure investment items. A technique (i.e., Fixed Effect Model) that is often used to solve knapsack problems, and the genetic algorithm are also used in this research to optimize the distribution. In the third year, the Taiwan』s infrastructure expenditures and IMD』s national competitiveness rankings will be used to compare the distribution suggested by the research and verify the prediction of the proposed model. The result can be used to help the Taiwan government plan the future infrastructure development.