A Comparison of Alternative Econometric Models of Iron and Steel Investment Behavior
比较了美国钢铁行业投资的几种计量经济模型，包括非线性新古典模型和线性新古典模型，基于1960-73年季度数据评估其表现，并预测1974-76年投资，帮助判断不同投资理论的适用性。
Numerous empirical studies have been undertaken in order to determine those factors that influence manufacturing investment in fixed capital (see Jorgenson, 1971). Most multi-industry studies employ quarterly data for a dozen or more two-digit industry groups. Results consistent across industries are difficult to achieve on account of the wide range of alternative specifications and selection of the best fitting among them. Comparison of results from different studies is further complicated by the use of different time periods, data sources, lag structures, and so on, or by the absence of a post-data model evaluation. Systematic comparisons are found in only a relative handful of empirical studies (see Jorgenson, Hunter, and Nadiri, 1970a, 1970b). This paper presents results for several alternative econometric models of investment in the U.S. iron and steel industry. Two general specifications are dealt with here: first, a nonlinear-neoclassical (N-N) model incorporating a constant elasticity of substitution (CES) production function and, second, a linear-neoclassical (L-N) model incorporating a liquidity variable and a non-specific production function. The empirical results provide a means of discriminating among competing hypotheses on the basis of their performance. However, we avoid unlimited data mining and, consequently, do not explore every possible specification for each model. The paper is divided into three sections. The first section discusses the investment models and presents empirical results based on quarterly data for the period 1960-73. Section II presents forecasts for 1974-76, and section III summarizes the main results from the study and comments on the investment function controversy. Data sources and variable definitions are contained in the appendix. I. The Investment Models and Empirical Results