Variable Returns to Scale in Production and Patterns of Specialization
分析一个行业规模报酬递增、另一个递减的两部门模型，发现小国完全专业化时倾向递减报酬行业，且福利最大化要求补贴递增行业、收缩递减行业，两国贸易均衡中递增报酬出口国可能不完全专业化。
This paper discusses some trade and welfare implications of the two-sector model based on increasing returns to scale (IRS) in one industry and decreasing returns to scale (DRS) in the other. It has been shown by Horst Herberg and Murray Kemp that, given homothetic production functions with IRS in one industry and DRS in the other, the production-possibilities frontier (PPF) is strictly concave to the origin near the IRS axis and strictly convex to the origin near the DRS axis. While this result constitutes an important finding and is contrary to the general impression among economists, even twelve years after the publication of Herberg and Kemp's paper, no attempt has been made to analyze its implications for welfare and patterns of specialization.' In this paper, I consider a simple twocommodity model with IRS in one industry and DRS in the other, and discuss some interesting implications of variable returns to scale (VRS). First, I demonstrate that, in this model, if a small, open economy specializes completely in production, it will do so in the DRS commodity and not in the IRS commodity as is generally believed. Second, if, at a given price ratio, an internal production equilibrium exists, a welfaremaximizing small, open economy will never specialize completely in production even though the PPF exhibits decreasing opportunity costs over a part of its range. Third, given output-generated economies and diseconomies of scale, welfare maximization for a small country requires a permanent tax subsidy scheme encouraging expansion of the IRS industry and contraction of the DRS industry. Finally, in a two-country model with identical tastes and technology across the countries, free trade equilibrium may result in incomplete specialization by the exporter of the IRS commodity and complete specialization by the exporter of the DRS commodity. In the course of the analysis, I argue that Tinbergen's formulation of Frank Graham's case for protecting the IRS industry is incorrect. I then present a possible reformulation of it. In Section I, the basic production model is introduced; in Section II, the model is analyzed; and in Section III, it is argued that all the results continue to hold in the multifactor case.