但按增加值计算，则两国的贸易逆差远比按出口总额计算为小.pptVIP

* * * * * 基于不同企业类型和不同贸易方式的面板数据建模，可以得到同时关于贸易总量与贸易结构的预测数据，二者互为调整依据。 * * 事实上[WY]I就是其它N个地区，在我后面研究的问题中就是３１个省市自治区对第I个地区共同影响，具体说来也就是与该省市出口共同波动的多个地区的出口值的一个汇总序列 * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 预测结果比较令人满意. 在对2005 年1 月至2005 年12 月的预测中,出口预 测最大误差率未超过4 % ,92 %的预测误差率在3 %以下,50 %的预测误差率在1 %以下. 进口预测最大误 差率为613 % ,50 %的预测误差率在3 %以下,42 %的预测误差率在2 %以下. 实际预测区间为18 个月,其 中前6 个月为测试区间,预测区间较长但误差率没有明显增大,尤其是出口预测,这说明预测模型的稳定 性较好. 第4 期一种新的集成预测方法———GPVECM 111 ① ② GPLAB 由Sara Silva 博士开发,是Matlab 的一个工具包. 我们的数据只能获取到2004 年6 月,所以无法做更新的实证研究. ? 1994-2009 China Academic Journal Electronic Publishing House. All rights reserved. * Model Leading Indicator Approach(1) Leading indicators Coincident indicators Lag indicators Industrial added value M1 Crude steel production Steel production Consumer confidence index released by America Conference Board OECD leading index of Russia OECD leading index of USA OECD leading index of Japan OECD leading index of Korea Export Import Processing trade export America overall industrial output Total retail sales of consumer goods Industrial products inventory America export price index Japan export price index Selecting leading indicators of export or import to construct leading index. Method: NBER business cycle analysis methods Model Leading indicator approach Leading index Peak: -16(mean),2.83(std);Valley: -5.67(mean), 3.51(std). Lag index Peak: 6.5(mean),4.95(std);Valley: 2(mean), 6.56(std). Forecasting foreign trade (4)： Policy analysis Aims of policy analysis Help improve the accuracy of forecasting. Models based on past correlations between macroeconomic variables can only handle forecasting without any changing of the institutional framework. These correlations can be expected to change when new policies are introduced. Help make policy. Policy makers make decisions based on possible outcomes of policies. Past policy evaluations by CEFS RMB appreciation Before 2005/7 Before 2010/6