Statistical Analysis of Economic Indicators

本研究的重点是经济周期经济指标的最新发展。特别对近年来经济指标发展的统计和实证方面进行了分析。为此，从时间序列和计量经济模型的角度比较了几种新经济指标的预测精度。分析的指标有：(1)多元时间序列方差成分模型（MTV模型），(2)贝叶斯VAR模型，(3)Stock & Watoson实验吻合模型，(4)计量经济模型。在1970年以后的日本经济周期中，我们强调MTV模式是有效的。特别是将MTV模型与马尔可夫切换模型（MSM）相结合，将成为拐点预测的有效工具。（该实验结果在1994年经济周期会议上报道）。Stock & Watsonis的另一个重要的实验一致性指标从理论和转折点预测上也是有效的。虽然与两态MTV=MSM模型相比，该方法非常耗时，但随着微处理器的快速发展，这一缺陷将得到解决。我们还开发了中等规模的日本计量经济模型。该模型的特征是货币部门的延伸（具有家庭投资组合选择部门）和实物部门与货币部门之间的联系。（该模型的模拟和短期重点由1994年商业周期研讨会报道）。我还没有完成这个模型对经济周期事件概率的估计，但是这个研究将会继续。大部分研究结果将在不久的将来发表（题目：经济指标的统计方法）。

This research focused on the recent development of economic indicators of business cycles. Especially, the statistical and empirical aspect of the recent development of the economic indicators are analyzed. For this purpose, the predictive accuracy of the several new economic indicators are compared from the time series and econometric models points of view. The indicators which have been analyzed are : (1) Multivariate time series Variance Component Model (MTV model), (2) VAR (Baysian VAR) model, (3) Experimental coincident model by Stock & Watoson, (4) Econometric Model. In the Japanese business cycle after 1970, we emphasize MTV model is valid. Especially, It will be useful tool for the turning point prediction to link MTV model and Markov Switching Model(MSM). (This experimental results are reported in the conference of Buisiness cycle 1994). Another important experimental coincident indicator by Stock & Watsonis is also valid from the theoretical and turning point prediction. Though this method is very time consuming in the comparison with 2 state MTV=MSM model, this fault will be solved by the rapid development of micro processor. We also developed medium size Japanese econometric model. The characteristics of this model is the extension of monetary sector (which has the sector of portfolio selection by households) and the link between physical and monetary sector. (the simulation and the short term focuses by this model are reported by the seminar of Business cycle, 1994). I haven't completed the estimation of event probabilities of business cycles by this model, but this study study will be continued. Mos of the results by this will be published in the near future (title : The statistical methods of economic indicators.