本文旨在建立隨機邊界產出成長率函數，在變動規模報酬的假設下，探討起始所得水準、實體資本、金融發展、人力資本與勞動力對各國經濟成長之影響，進而重新檢視收斂假說，並進行國家技術效率與總要素生產力分解之跨國分析。蒐集1980 年至2000 年74個樣本國家之平衡縱橫資料，實證分析主要發現有四：(i) 實體資本與人力資本具有互補關係，金融發展與實體資本具替代性，但與人力資本呈互補關係；(ii) 低、中所得國家皆呈現所得收斂現象，且低所得國家之收斂速度較快，而高所得國家則無所得收斂情形；(iii)高所得國家享有較高的資本邊際生產力，低所得國家則以勞動與人力資本的邊際貢獻較大，並享有規模經濟；(iv) 全體國家之總要素生產力平均以每年3.52% 的速度成長，其中以中所得國家速度最快；中、高所得國家主要成長來源皆為技術進步，低所得國家則為規模效果。

This study applies the stochastic translog production frontier under variable returnsto scale to investigate the effects of initial income, capital stock, financial development,human capital, and labor on a country’s economic growth, in which each variable istransformed by taking the first difference in order to delete any inherent non-stationarity.This modeling tests the convergence hypothesis and decomposes productivity change intovarious elements. We collect data from 74 countries, covering 1980-2000, and findevidence that capital stock and human capital are complements, and that financialdevelopment is a substitute for human capital, but a complement for human capital. Thisstudy supports the convergence hypothesis through low and middle-income countries,whereby the speed of convergence in low-income countries is faster than in middleincomecountries, while this study rejects the hypothesis for high-income countries. Wenote that the marginal productivity of capital of high-income countries is the highest, andthat the marginal productivities of labor and human capital of low-income countries arethe highest. Moreover, the average growth rate of total factor productivity for thesample states is equal to 3.52% per annum.