# Simple returns in percentage

AXP
20 15 10 5 0 -5 -10 -15 -20 99 00 01 02 03 04 05 06 07 08

Log returns in percentage
LAXP
20 15 10 5 0 -5 -10 -15 -20 99 00 01 02 03 04 05 06 07 08

CAT
15 16 12 8 5 4 0 -4 -8 -10 -12 -16 99 00 01 02 03 04 05 06 07 08 99 00 01 02

LCAT

10

0

-5

-15

03

04

05

06

07

08

SBUX
20

LSBUX
20

10

10

0
0

-10
-10

-20
-20

-30
-30 99 00 01 02 03 04 05 06 07 08

-40 99 00 01 02 03 04 05 06 07 08

053 0.299 2083.701 12. Dev.043 1.494 7.337 -0.785 0 2515 0.459 11. Error t-statistic LAXP LCAT LSBUX -0.518 -28.735 13.015/[2.05 t-statistic (log variable) = mean/std error LAXP-log returns for American Express: t-statistics = -0. Dev.453/(2515)^0.316 2332.459 8. The transformation for log returns is generated from the eviews with laxp=log(axp1+1)*100. The descriptive statistics results for both simple and log returns are shown in tables below.453 2.316 Degree of freedom = N-1 =2514 .8 0 2515 0.84 0 2515 0. we then denote the simple returns expressed in percentage as axp.95 0 2515 0.082 Kurtosis 9. cat.171 2. axp1. Null Hypothesis H0: mean of log returns equals zero Alt.686 -33.221 Daily simple returns AXP CAT SBUX Mean 0. excess kurtosis is kurtosis minus 3 Testing hypothesis of zero expected mean To test if the mean of the log returns of each stock is zero.489 13. i. Hypothesis H1: mean of log returns is not zero We reject the null hypothesis if the t-statistic estimated falls beyond the 5 percent confidence limit. if p value is less than 0.494 4.595 -14. Skewness Kurtosis *Excess Kurtosis Maximum Minimum Jarque-Bera Probability Observations Std.659 -19.202 -0. lcat.99 0 2515 0.597 9.035 0. Error t-statistic 3842.701 15. and lsbux.375 8034.e.055 4.755 *Excess Kurtosis 6.e.755 Maximum 17.898 *Since kurtosis is 3. and sbux1 are the raw data for simple returns. Daily log returns Mean Std.036 0.012 2.1 From the Eviews workfile.5] = -0.830 17609.015 0. [i.352 -15.635 Minimum -17.015 0.908 6.723 14.048 Std.055 7.286 Jarque-Bera Probability Observations Std. we set our hypotheses. axp=axp1*100] The log transformed data for the simple returns in percentage is denoted with laxp.170 2.696 -0.446 2.060 0.249 4466.908 16. 2.043 0.049 -0.927 14. and sbux.012 -0.683 Skewness -0.Econ 604 Assignment 1 Omotola Awojobi (std number 116061) Solution for 1.779 0 2515 0.. cat1.049 0.054 0.

log returns for Starbucks: t-statistics = 0. t-statistic @5% level of significance and df 2514 is 1. the zero mean expectation holds for all the stocks.p-value = TDIST(0. Null Hypothesis H0: skewness (SLAXP) measure of the returns is zero Alt.96.96 and 1.696/(2515)0. two-tailed) = 0.000 From the t-table.892 p-value (skewness) = TDIST(6.752 imported from excel The result for p value is far greater than the 0.830.5] = -6.5] = 0. 2514. Hypothesis H1: skewness (SLAXP) measure of the returns is not zero We reject the null hypothesis if the t-statistic estimated falls beyond the 5 percent confidence limit. this implies that the hypothesis of zero expected return cannot be rejected at the 5 percent confidence interval.05 t-statistic (skewness) = SLAXP/[(6/T)0. We conclude that.5] = -0. 2514.171/(2515)0.316. 2514.337/[(6/2515)0.012/[2.892.05 confidence limit. This implies that the price of these stocks have not significantly changed Solution for 1. We test if skewness of the log returns for American express stock is zero.830 Degree of freedom = N-1 =2514 p-value = TDIST(0.4 Testing hypothesis for skewness and excess kurtosis LAXP a. if tprobability is less than 0.e. 2514. two-tailed) = 0.96. two-tailed) = 0.221 Degree of freedom = N-1 =2514 p-value = TDIST(0. two-tailed) = 0. LCAT . result shows that the hypothesis of zero expected return cannot be rejected at the 5 percent confidence interval.log returns for Caterpillar: t-statistics = 0. we reject the null hypothesis and conclude that skewness measure of log .825 imported from excel For Starbucks stock as well.5] = 0. i. LSBUX . based on the sample data applied for this exercise.036/[2. Since t-statistic (skewness) falls beyond the range of -1.221.406 imported from excel The result shows that the hypothesis of zero expected return cannot be rejected for CAT stocks at the 5 percent confidence interval.

809502 7.5] = 6.06983 4.239039 0. Hypothesis H1: excess kurtosis (KLAXP-3) measure of the returns is not zero We reject the null hypothesis if the t-statistic estimated falls beyond the 5 percent confidence limit.126013 -0.5 a.208 0 2322 2. we reject the null hypothesis and conclude that excess kurtosis of log returns distribution for American Express is significantly different from zero. LPC_CAD LPC_EURO LPC_POUND LPC_YEN -0.653943 0. b.587648 0.186 0 2322 4.663232 -0. Skewness Kurtosis Excess Kurtosis Maximum Minimum Jarque-Bera Probability Observations .01099 -0.25503 5. Dev. 2514.96625 4896.809502 10.494/[(24/2515)0. b.returns distribution for American Express is significantly different from zero. Test for excess kurtosis of the log returns for American express stock.5] = 66.587413 3.395083 -0. PC_CAD is for single period percentage change in exchange rate generated with genr pc_cad=(cad1/cad1(-1)-1)*100 from eviews. two-tailed) = 0.311 0 2322 Mean Std. LPC_CAD is for log returns in percentage.06983 7.620792 -3.000 Since p value (kurtosis) is far below the 5% significance.005747 -0.215648 2210. See eviews workfile title “Tola_604wf1_5”.8224 0 2322 4. Solution for 1.67105 11.003101 769. if p value is less than 0.05 t-statistic (excess kurtosis) = (KLAXP-3)/[(24/T)0.708365 -5.478.25503 2. i.071599 6615.617662 0.587413 8. Null Hypothesis H0: excess kurtosis (KLAXP-3) measure of the returns is zero Alt.002185 0.006819 0. The log returns are significantly skewed to the left. same applies to other currencies.478 p-value (excess kurtosis) = TDIST(66.434858 -4. generated with genr lpc_cad=log((pc_cad/100)+1)*100.806962 -5. The log returns got fat tail.e.

they all have fat tail with the CAD pair showing stronger indication of volatility.2011074 45.5399634 LPC_YEN -0.70245064 2.478967502 -7. Density plot for daily log returns dollar to euro .c.15875122 -13. d.63470497 69.19777761 27.1225181 t-statistic (mean) t-statistic (skewness) t-statistic (excess kurtosis) The mean log returns is statistically proven to be zero for all currencies which means they did not change much over the period probably only gained value marginally. all other distributions are significantly skewed to the left. Trying a t-statistic for their significance. From the results presented in table above. LPC_CAD LPC_EURO LPC_POUND -0.448354385 -4.772197453 81. Except from the US dollar pair with euro.55915799 0. we have the table below. Also.809820758 -0. mean log returns on all the exchange pairs look very close to zero and the standard deviation is not so clear to tell if they are statistically different from zero.