金融时间序列计算分析题1

计算分析题1

Consider the daily simple returns of American Express (AXP), CRSP value-weighted index (VW), CRSP equal-weighted index (EW), and the S&P composite index (SP) from September 01, 2001 to September 30, 2011. Returns of indices include dividends. The data are in the file d-axp3dx-0111.txt (date, axp, vw, ew, sp).

(a) Compute the sample mean, standard deviation, skewness, excess kurtosis,minimum, and maximum of each simple return series.

结果：

code	mean	standard deviation	minmum	maxmum	skew	excess kurtosis
axp	0.000534	0.026368407	-0.17595	0.206485	0.459773	9.592052847
vw	0.000224	0.013651835	-0.08976	0.114889	-0.09832	7.982133846
ew	0.000626	0.012080369	-0.07824	0.107422	-0.24741	8.108427735
sp	9.42E-05	0.013779117	-0.09035	0.1158	0.008152	8.532667329

各个简单收益率都为尖顶峰，axp呈明显正偏态，ew呈明显负偏态，vw和sp没有明显的偏态。

(b) Transform the simple returns to log returns. Compute the sample mean,standard deviation, skewness, excess kurtosis, minimum, and maximum of each log return series.
$$对数收益率：r_t=log\frac{P_t}{P_{t-1}}=log(1+R_t)R_t为简单收益率$$

结果：

code	mean	standard deviation	min	max	skew	excess kurtosis
axp	0.000188	0.026294	-0.19352	0.187711	0.020992	9.020499
vw	0.000131	0.01367	-0.09405	0.108755	-0.30035	7.880082
ew	0.000553	0.0121	-0.08147	0.102035	-0.42732	8.017712
sp	-7.5E-07	0.01379	-0.0947	0.109572	-0.20636	8.322826

各个对数收益率都为尖顶峰，vw、ew、sp呈明显负偏态，axp没有明显的偏态。

© Test the null hypothesis that the mean of the log returns of AXP stock is zero. Use 5% significance level to draw your conclusion.

答：通过t检验和z检验，AXP的对数收益率在5%显著性水平的双边检验下不为0。

t检验结果图：

z检验结果图：

代码附录：

#导入d-axp3dx-0111.txt文件
data1 <- read.table(
  "D:/Rwork/金融时间序列分析数据/d-axp3dx-0111.txt",
  header=T,stringsAsFactors = F)
#(a) Compute the sample mean, 
#standard deviation, skewness, excess kurtosis,minimum, 
#and maximum of each simple return series. 
library(psych)
myvars <- c("axp","vw","ew","sp")
resulta <- describe(data1[myvars])
#导出结果
library(xlsx)
write.xlsx(resulta,"resulta.xlsx")

#(b) Transform the simple returns to log returns.
#Compute the sample mean,standard deviation, 
#skewness, excess kurtosis, minimum, and maximum of
#each log return series.
data2 <- transform(data1,axp=log(1+axp),vw=log(1+vw),
                   ew=log(1+ew),sp=log(1+sp))
resultb <- describe(data2[myvars])
write.xlsx(resultb,"resultb.xlsx")
#Test the null hypothesis that the mean of the log returns of AXP  stock is zero. 
#Use 5% significance level to draw your conclusion.
library(BSDA)
t.test(x=data2["axp"],mu=0,sigma.x=0.05,alternative="two.sided")
z.test(x=data2["axp"],mu=0,sigma.x=0.05,alternative="two.sided")