证券投资分析作业
CAPM模型在中国资本市场的有效性检验
1、数据选取
此次实验主要考察CAPM模型在中国电力行业是否适用，因此随机抽取了电力行业的十只股票（时间段为2010年1月1日—2010年12月31日），分别为
股票代码 股票简称 股票代码 股票简称 002039 黔源电力 600101 明星电力 600116 三峡水利 600292 九龙电力 600310 桂东电力 600452 涪陵电力 600505 西昌电力 6004 乐山电力 600674 川投能源 600969 郴电国际
选取沪深300指数为综合指数，选取2010年的国债的利率作为无风险资产的收益率（0.025）。
2、β系数的确定
CAPM模型中，β系数可以表述为：Ri – Rf =αi + βi( Rm - Rf) + εi，其中Ri为每一种证券的收益率，Rf为无风险收益率，Rm为市场收益率。
使用Eviews软件对每只股票每日风险溢价与市场组合风险溢价进行回归，得到每只股票的β值。如下：
（1）黔源电力
Dependent Variable: Y
Method: Least Squares
Date: 12/26/11 Time: 16:35
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.008685
0.002294 -3.786006
0.0002
X 0.616613 0.076324 8.078883 0.0000
????Mean dependent
R-squared
0.214509 var
-0.024413
Adjusted R-squared
????S.D. dependent
0.211223 var
0.021210
S.E. of regression
????Akaike info
0.018838 criterion
-5.097652
????Schwarz
Sum squared resid 0.084811 criterion
-5.068732
Log likelihood 616.2670 ????F-statistic 65.26835
Durbin-Watson stat
????Prob(F-1.914885 statistic)
0.000000
（2）明星电力
Dependent Variable: Y2
Method: Least Squares
Date: 12/26/11 Time: 16:46
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.032526
0.007661 -4.245595
0.0000
-X
0.215975
0.2542 -0.847320
0.3977
R-squared 0.002995
????Mean dependent
-
var 0.027017
Adjusted R-squared
-????S.D. dependent
0.001177 var
0.062873
S.E. of regression
????Akaike info
0.062910 criterion
-2.685947
????Schwarz
Sum squared resid 0.9454 criterion
-2.657027
Log likelihood 325.6566 ????F-statistic 0.717951
Durbin-Watson stat
????Prob(F-1.196603 statistic)
0.397665
（3）三峡水利
Dependent Variable: Y3
Method: Least Squares
Date: 12/26/11 Time: 16:48
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.029398
0.0042 -6.853614
0.0000
-X
0.160104
0.142712 -1.121869
0.2630
????Mean dependent
R-squared
0.005238 var
-0.025314
Adjusted R-squared
????S.D. dependent
0.001076 var
0.035242
S.E. of regression
????Akaike info
0.035223 criterion
-3.845971
????Schwarz
Sum squared resid 0.296518 criterion
-3.817051
Log likelihood 465.4395 ????F-statistic 1.258591
Durbin-Watson stat
????Prob(F-1.523152 statistic)
0.263044
（4）九龙电力
Dependent Variable: Y4
Method: Least Squares
Date: 12/26/11 Time: 16:50
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.023708
0.004362 -5.434675
0.0000
-X
0.003584
0.145136 -0.024693
0.9803
????Mean dependent
R-squared
0.000003 var
-0.023616
Adjusted R-squared
-????S.D. dependent
0.004182 var
0.035747
S.E. of regression
????Akaike info
0.035821 criterion
-3.812283
????Schwarz
Sum squared resid 0.306677 criterion
-3.783363
Log likelihood 461.3801 ????F-statistic 0.000610
Durbin-Watson stat
????Prob(F-1.598474 statistic)
0.980321
（5）桂东电力
Dependent Variable: Y5
Method: Least Squares
Date: 12/26/11 Time: 16:52
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.027401
0.003728 -7.351010
0.0000
-X
0.174539
0.124019 -1.407360
0.1606
????Mean dependent
R-squared
0.008219 var
-0.022949
Adjusted R-squared
????S.D. dependent
0.004069 var
0.030672
S.E. of regression
????Akaike info
0.030609 criterion
-4.126758
????Schwarz
Sum squared resid 0.223927 criterion
-4.097838
Log likelihood 499.2743 ????F-statistic 1.980662
Durbin-Watson stat
????Prob(F-1.567083 statistic)
0.160620
（6）涪陵电力
Dependent Variable: Y6
Method: Least Squares
Date: 12/26/11 Time: 16:53
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.027569
0.009995 -2.758287
0.0063
X 0.028673 0.332537 0.086226 0.9314
????Mean dependent
R-squared
0.000031 var
-0.028300
Adjusted R-squared
-????S.D. dependent
0.004153 var
0.081904
S.E. of regression
????Akaike info
0.082074 criterion
-2.154127
????Schwarz
Sum squared resid 1.609937 criterion
-2.125208
Log likelihood 261.5723 ????F-statistic 0.007435
Durbin-Watson stat
????Prob(F-1.109620 statistic)
0.931359
（7）西昌电力
Dependent Variable: Y7
Method: Least Squares
Date: 12/26/11 Time: 16:55
Sample: 1 241
Included observations: 241
Variable
Coeffici
Std. Error
t-
Prob.??
ent Statistic
-C
0.0234
0.004241 -6.233043
0.0000
X 0.016241 0.141098 0.115107 0.9085
????Mean dependent
R-squared
0.000055 var
-0.026848
Adjusted R-squared
-????S.D. dependent
0.004128 var
0.034753
S.E. of regression
????Akaike info
0.034825 criterion
-3.868717
????Schwarz
Sum squared resid 0.2849 criterion
-3.839798
Log likelihood 468.1804 ????F-statistic 0.013250
1.452457
Durbin-Watson
????Prob(F-
0.908457
stat statistic)
（8）乐山电力
Dependent Variable: Y8
Method: Least Squares
Date: 12/26/11 Time: 16:56
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.028174
0.0039 -7.107256
0.0000
X
-
0.131888 -1.303503 0.1937
0.171916
????Mean dependent
R-squared
0.007059 var
-0.0237
Adjusted R-squared
????S.D. dependent
0.002905 var
0.032599
S.E. of regression
????Akaike info
0.032552 criterion
-4.003721
????Schwarz
Sum squared resid 0.253245 criterion
-3.974802
Log likelihood 484.4484 ????F-statistic 1.699119
Durbin-Watson stat
????Prob(F-1.733619 statistic)
0.193657
（9）川投能源
Dependent Variable: Y9
Method: Least Squares
Date: 12/26/11 Time: 16:58
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.028579
0.003039 -9.402725
0.0000
-X
0.144156
0.101126 -1.425514
0.1553
????Mean dependent
R-squared
0.008431 var
-0.024902
Adjusted R-squared
????S.D. dependent
0.004282 var
0.025013
S.E. of regression
????Akaike info
0.024959 criterion
-4.534903
????Schwarz
Sum squared resid 0.148885 criterion
-4.505984
Log likelihood 548.4558 ????F-statistic 2.032090
Durbin-Watson stat
????Prob(F-1.710352 statistic)
0.155313
（10）郴电国际
Dependent Variable: Y10
Method: Least Squares
Date: 12/26/11 Time: 16:59
Sample: 1 241
Included observations: 241
Coeffici
Variable
t-
ent Std. Error Statistic Prob.??
-C
0.022969
0.003915 -5.866217
0.0000
X 0.072408 0.130268 0.555835 0.5788
????Mean dependent
R-squared
0.001291 var
-0.024816
Adjusted R-squared
-????S.D. dependent
0.002888 var
0.032105
S.E. of regression
????Akaike info
0.032152 criterion
-4.028440
Sum squared resid 0.247062
????Schwarz
-
criterion 3.999520
Log likelihood 487.4270 ????F-statistic 0.3052
Durbin-Watson stat
????Prob(F-1.756510 statistic)
0.578844
3、用求出的10只股票的β值与十只股票的平均收益率进行回归，如下：
Dependent Variable: YY
Method: Least Squares
Date: 12/26/11 Time: 17:27
Sample: 1 10
Included observations: 10
Variable
Coeffici
Std. Error
t-
Prob.??
ent Statistic
-5.47E-C
05
0.000603 -0.090685
0.9300
XX 1.30E-05 0.002598 0.005022 0.9961
????Mean dependent
R-squared
0.000003 var
-5.49E-05
Adjusted R-squared
-????S.D. dependent
0.124996 var
0.001796
S.E. of regression
????Akaike info
0.001905 criterion
-9.511885
????Schwarz
Sum squared resid 2.90E-05 criterion
-9.451368
Log likelihood 49.55942 ????F-statistic 2.52E-05
2.042840
Durbin-Watson
????Prob(F-
0.996116
stat statistic)
即样本回归方程为
Yt = -5.47 E-05 + 1.30 E-05 +εi
4、统计检验
r2 = 0.000003，说明仅有总离差平方和的0.003%被样本回归直线解释，回归直线对样本点的拟合优度非常低。
给出显着性水平α=0.05，P>α，t检验不能通过；F检验也不能通过。
从以上的检验可以看出，此模型没有通过各种检验，拟合不好，不能代表x与y的关系。
5、结论
通过分析可以看出，CAPM模型对我国资本市场上的电力行业不适用，通过更多的分析可以得出，CAPM模型对我国资本市场是无效的。
我国资本市场是导向型市场，采用核准制度，是计划经济的产物，资本市场还没有实现市场完全控制，资本未达到自由流动，还存在信息不对称、经济发展程度落后于发达国家、国际金融环境恶化等现象，加之CAPM模型的假设条件比较苛刻，因此在中国资本市场上应用这一模型极为困难。
