Impact of crisis periods

larger economic crises labelled in Figure 5.5 Section 5.2.2.

2006-2013

In the previous analysis of VIX, the mean returns displayed a substantial im-provement of the 1-month rebalancing strategies over this sub-period. This improvement was even larger for VXY, as can be seen in Table 5.9. In fact, the mean return of Reverse1M1C increased from 8.92% to 11.42% after apply-ing the VXY signal instead of VIX to time the strategy. The correspondapply-ing static carry trade, Carry1M1C, generated a mean return 1.40%, indicating sub-stantial improvement over this sub-period. Contrary to the static carry trade, all the mean returns obtained from the 1-month rebalance strategies measures were statistically significant at a 5% significance level. While the 3-month re-balance strategies improved compared to the static carry trade as well, these were far from the levels observed for the 1-month strategies. Reverse3M1C increased from -2.65% to 3.14% after applying the VXY signal instead of VIX to time the strategy. Both signals identified a crisis period in October 2008, where the monthly return of Carry1M1C in the subsequent month was down to -10.33%. However, in January 2009, the carry1M1C generated a negative return of -7.29%, which was only identified by VXY. This might partly explain the fairly larger improvement compared to VIX.

2013-2020

As described in Section 5.3.1, this period is characterised by a low-interest rate environment leading to small interest rate differentials, which is also reflected in the mean returns using VXY. The performance measure of 1-month re-balance strategies economically improved upon the corresponding static carry trade strategies, but the levels are far from the ones observed in the previous sub-periods. The greatest improvement is observed for Reverse1M1C, which increased from 0.98% to 3.49%. Although this period include the US-China trade war and COVID-19 crises, the impact is inherently small as the negative and positive returns observed over the period is small.

2007-2009 and the low-interest rates diminishing towards zero after the crisis.

After applying the dynamic strategies, the mean returns improved predomi-nantly for those rebalanced monthly. For this reason, the analysis will mainly focus on the 1-month rebalance strategies throughout the remaining parts of the analysis. To elucidate the impact of the crisis periods independently, the mean returns of Carry1M1C, Exit1M1C and Reverse1M1C are decomposed into the respective crisis period in this section.

VIX signal

Table 5.10 presents the mean return of Carry1M1C along with the impact of applying Exit1M1C and Reverse1M1C based on VIX of each crisis period.

Overall, the impact of using VIX increased the mean return of the Carry1M1C with 1.311% and 2.621% after applying the dynamic strategies. However, most of it can be contributed to the impact found during the financial crisis, which led to an improvement of the mean return of 1.338% and 2.677% respectively.

This reflects the greater potential for improvement as the losses which occurred during this period were higher, compared to the other crisis periods. During the remaining crisis periods, the potential for improvement was smaller, which is also reflected by the mean returns being well below 1%. During the Dot-com-bubble peaks, the Greek crisis and US-china trade war, the impact was even negative. The three crisis periods are characterised as having implied volatility levels that never increased to the highest levels observed in the periods around the Lehman collapse, Greek crisis and COVID-19. Disregarding the positive impact during the financial crisis, the total impact is negative. Therefore, re-lying on the signal may lead to even worse losses to carry investors over some crisis periods. It is important to note that, the signal is based on a rolling window, and therefore take into account, that an increase in volatility may not be perceived as material as the same increase during other periods. In hindsight, defining these as crisis periods could possibly have been avoided, by selecting different model parameters, including an even longer rolling window.

Table (5.10) Impact of the VIX signal

This table reports the mean return of Carry1M1C along with the impact of using the VIX signal in the dynamic strategies, Exit1M1C and Reverse1M1C over different crisis periods defined by VIXt−1. Other crisis periods, represents periods which occurred on months outside the range of dates of the labelled crisis periods.

Crisis period Start date End date Carry1M1C Exit1M1C Reverse1M1C Dot-com bubble peaks 30/03/2001 30/09/2002 0.335% -0.335% -0.670%

Financial crisis 28/02/2007 28/11/2008 -1.338% 1.338% 2.677%

Greek crisis 31/08/2011 30/09/2011 0.070% -0.070% -0.139%

Brexit 30/06/2015 30/09/2015 -0.169% 0.169% 0.337%

US-China trade war 31/10/2018 31/12/2018 0.105% -0.105% -0.211%

COVID-19 crisis 28/02/2020 31/03/2020 -0.013% 0.013% 0.026%

Other crisis periods -0.301% 0.301% 0.601%

Total impact -1.311% 1.311% 2.621%

VXY signal

Table 5.10 presents the mean return of Carry1M1C along with the impact, applying Exit1M1C and Reverse1M1C based on VXY of each crisis period.

Comparing this to Table 5.10, the impact obtained from using the signal based on VXY is overall higher. As already discussed in the preceding sections, VIX and VXY identified the same crisis periods 14 times over the sample period.

During the period, defined as the financial crisis, the VXY signal identified three more crisis periods compared to VIX, leading to an increase in the mean return over the financial crisis of 1.241% points. Both signals identified crisis periods which were able to make substantial improvements on the 1-month rebalancing periods during the financial crisis. The financial crisis alone led to an increase in the mean return of the Reverse1M1C of 2.677% points compared to the carry trade. Similarly to VIX, is the impact of the remaining crisis periods, negative in two crisis periods. However, the magnitude of the impact of the two periods is lower than the positive impact of the other periods. And even if the financial crisis were excluded, the overall impact would still be positive.

Overall, the impact of using VXY increased the mean return of the Carry1M1C with 2.193%- and 4.386% points after applying the exit- and reverse strategies.

This, along with the ability to identify months of highly negative returns, the VXY demonstrates itself as a good crisis indicator.

Table (5.11) Impact of the VXY signal

This table reports the mean return of Carry1M1C along with the impact of using the VIX signal in the dynamic strategies, Exit1M1C and Reverse1M1C over different crisis periods defined by VXYt−1. Other crisis periods, represents periods which occurred on months outside the range of dates of the labelled crisis periods

Crisis period Start date End date Carry1M1C Exit1M1C Reverse1M1C Financial crisis 31/07/2008 30/01/2009 -1.599% 1.599% 3.198%

Greek crisis 30/09/2011 31/12/2014 -0.128% 0.128% 0.256%

Brexit 31/03/2015 29/02/2016 0.032% -0.032% -0.064%

US-China trade war 31/12/2018 31/12/2018 -0.048% 0.048% 0.095%

COVID-19 crisis 31/03/2020 31/03/2020 0.009% -0.009% -0.017%

Other crisis periods -0.459% 0.459% 0.918%

-2.193% 2.193% 4.386%