Normalized cointegrating coefficients (standard error in parentheses)    
       LFR        LIG              LFE            C 
 1.000000 -0.548610  0.129287  0.538067 
  (0.10004)  (0.04361)  (0.11019) 
    
上面是通过Johansen协整检验得到的协整方程,LFR与LIG成正相关关系,若对这三个变量进行线性回归的时候得到下面的结果:

Dependent Variable: LIG 
Method: Least Squares 
Date: 10/08/08   Time: 20:30 
Sample: 1985 2006 
Included observations: 22    
    
Variable      Coefficient          Std. Error         t-Statistic           Prob. 
    
C                1.133773             0.027333         41.47944        0.0000
LFR          -0.119376             0.163022         -0.732269       0.4729
LFE           0.383898              0.043638          8.797288       0.0000
    
R-squared 0.808403     Mean dependent var  0.933286
Adjusted R-squared 0.788235     S.D. dependent var  0.158785
S.E. of regression 0.073069     Akaike info criterion  -2.268691
Sum squared resid 0.101444     Schwarz criterion  -2.119912
Log likelihood 27.95560     F-statistic  40.08328
Durbin-Watson stat 0.605830     Prob(F-statistic)  0.000000
    
从上面的回归结果来看,似乎LIG 与LFR成负相关呢?这是什么原因?越来越糊涂了!高人指点!