41
42
Table 13: The excess return on the Fama/French Europe index during the period of 24th of February 1999 to 31st of December 2019 using different trading strategies.
Average annual excess return
(AAER)
Standard deviation of annual returns
(SDAR)
Sharpe ratio (SR)
Value of $1 invested on 24th of February
1999 ($1 invested) Holding the index on all
days 6.35% 22.90% 0.28 $3.135
Holding the index only in
weeks 0,2,4,6 0.95% 13.73% 0.07 $1.185
Holding the index only in
weeks -1,1,3,5 4.73% 14.64% 0.32 $2.646
Holding the index only in
week 3 2.18% 4.29% 0.51 $1.483
4.6.1 Comparison of investment strategies between periods
Table 14 shows comparison of the investment strategies between the period of 1999-2014 when monetary policy decision was made every two weeks and since 2015 when there is a monetary policy decision every six weeks.
The returns of the Fama/French Europe index are slightly lower for the period of 1999-2014 with a higher standard deviation, resulting in a considerably lower Sharpe ratio when the index is held on all days over the period of 1999-2014. The biweekly AAER are higher when the index is held in odd weeks over the period of 1999-2014, whereas for the period of 2015-2019 the returns are higher when the index is held in even weeks.
By holding the index in week 3 over the period of 1999-2014 the AAER is 3.11% over the period with a standard deviation of 4.24% so the Sharpe ratio for the strategy is 0.73. By using the same strategy over the period of 2015-2019 the AAER is -0.24% and the standard deviation of returns 3.76%, giving a Sharpe ratio of -0.07.
43
Table 14: Comparison of the excess return on the Fama/French Europe index using different investment strategies between the periods of 1999-2014 and 2015-2019.
Return of investment strategies for the period of 1999-2014
AAER SDAR Sharpe Ratio
Holding the index on all days 6.23% 24.72% 0.25
Holding the index only in weeks 0,2,4,6 0.25% 15.05% 0.02
Holding the index only in weeks -1,1,3,5 5.26% 15.80% 0.33
Holding the index only in week 3 3.11% 4.24% 0.73
Return of investment strategies for the period of 2015-2019
AAER SDAR Sharpe Ratio
Holding the index on all days 7.20% 17.96% 0.40
Holding the index only in weeks 0,2,4,6 4.12% 7.77% 0.53
Holding the index only in weeks -1,1,3,5 2.53% 11.46% 0.22
Holding the index only in week 3 -0.24% 3.76% -0.07
44
5 Conclusion
In this thesis the returns of the Fama/French Europe index over a six-week period following a monetary policy decision by the ECB were investigated. Focus was set on even weeks as Cieslak et al. (2019) had showed that average five-day excess returns on the CRSP value weighted index over 1-month US Treasury Bill rate was exclusively positive during even weeks following a Federal Open Market Committee meeting. Two studies were performed on the FFE index.
First, an event study using the methodology of MacKinlay (1997) where the MSCI ACWI ex EMU Index was used for estimation of normal returns of the FFE index. Two types of tests were performed on the abnormal returns, a parametric and a non-parametric test. The parametric test performed was the standard t-test while the generalized rank test was the nonparametric test.
Second, the excess returns of the FFE index were studied using the methodology by Cieslak et al. (2019) where the 1-month U.S. T-bill rate was used for calculation of excess returns. The average five-day excess return was studied as well as the relationship between monetary policy decisions and weekly returns through an ordinary least squares regression.
In order to investigate the returns of the FFE index following a monetary policy decision, four hypotheses were put forth. The results gathered through the use of event studies and displayed in Table 3 show that we
fail to reject 𝐻0 that 𝐶𝐴𝑅̅̅̅̅̅̅(𝑤𝑒𝑒𝑘0) = 0 fail to reject 𝐻1that 𝐶𝐴𝑅̅̅̅̅̅̅(𝑤𝑒𝑒𝑘2) = 0 fail to reject 𝐻2 that 𝐶𝐴𝑅̅̅̅̅̅̅(𝑤𝑒𝑒𝑘4) = 0 fail to reject 𝐻3 that 𝐶𝐴𝑅̅̅̅̅̅̅(𝑤𝑒𝑒𝑘6) = 0.
The hypotheses tests show that the returns of the FFE index are not significantly different than the returns predicted by the market model during even weeks.
The replication study does not show a clear sign of a biweekly pattern of returns in the weeks after a monetary policy decision by the ECB. Even though there are changes in the five-day average excess return throughout the weeks as seen in Figure 2, they do not imply a weekly
45 pattern. The regression analysis between the FFE index excess return and dummy variables on even and odd weeks showed some indication that returns were negative during even weeks and positive during odd weeks, but none of the beta coefficients were significantly different from zero.
The event study results from the comparison between the period of 1999-2014 and 2015-2019 showed that abnormal returns were not signifcantly different from zero or differed significantly from the mean for any of the weekly periods.
The investment strategy analysis indicates that returns are higher during odd weeks than even weeks following a monetary policy decision by the ECB. While both returns are positive, the average annual excess return on the FFE index was 4.73% on odd weeks whereas on even weeks the AAER was 0.95%. As returns were positive for both odd weeks and even weeks, the highest AAER was achieved by holding the FFE index on all weeks. However, the Sharpe ratio was higher when the index was only held on odd weeks.
When comparing the investment strategies between the 1999-2014 and 2015-2019 period the odd week returns are higher compared to the even week returns over the period of 1999-2014. During the period of 2015-2019 the reverse is true, as the even week returns are higher than the odd week returns.
The analysis of the average five-day excess return did not show an indication of a biweekly pattern for the 1999-2014 or the 2015-2019 period. The regression analysis on dummy variables on even and odd weeks did not show a relationship between monetary policy decision and weekly returns. The only beta coefficient that was significantly different from zero was on week 3 during the period of 1999-2014 where the beta coefficient was positive at the 5% significance level.
Summing up the results of the analyses performed, the conclusion of this research is that there is not compelling evidence that the Fama/French Europe index follows a biweekly pattern following a monetary policy decision by the ECB over the period of 1999-2019. The results show some indication that returns are lower during even weeks than during odd weeks. The beta coefficient for even weeks in Table 5 was negative while the beta coefficient was postive for odd weeks in Table 6. None of the beta coefficient in Table 5 and 6 did
46 however differ significantly from zero. The return was also higher when the FFE index was only held on odd weeks compared to even weeks as can be seen in Table 13.
Looking at the returns of the FFE index between the periods of 1999-2014 and 2015-2019 there is an indication that the excess returns are lower on even weeks and higher on odd weeks over the period of 1999-2014 and vice versa over the period of 2015-2019 as can be seen in Table 9, Table 10, and Table 14. The evidence are however not strong enough to be statistically significant.