Decomposing volatility

Transcription

1 MARKET RESEARCH Decomposing volatility FTSE EPRA/NAREIT Developed Europe Index S e p t e m b e r 217

2 Content table Executive Summary 3 Indices & stocks: Volatility evolution and key facts 3 Data and Methodoogy 8 Volatility decomposition and long-term determinants 9 Conclusions 15 Appendix 16 Ali Zaidi Indices & Research Director David Moreno Indices & Research Analyst Any interpretation and implementation resulting from the data and finding within remain the responsibility of the company concerned. There can be no republishing of this paper without the express permission from EPRA. Inna Maslova Indices & Research Analyst Pol Camps Indices & Research Analyst

3 I. EXECUTIVE SUMMARY Many listed real estate companies in Europe are included as constituents of various equity market indices and in addition to dedicated real estate indices, which contributes to the volatility in real estate stocks across the continent. In addition, the FTSE EPRA/NAREIT Developed Europe TR Index (RPRA), widely used as an industry and regional benchmark, is based in EUR, then returns of companies based on countries out of the Eurozone are also influenced by the dynamics of the local currency against the euro. This report aims to analyse and understand the dynamics of the individual and aggregated volatility of the RPRA constituents. To do so, the volatility of a selection of these constituents is decomposed into the following four main sources: the company s own dynamics, listed real estate continental and sectoral indices, the local equity markets, and the currency movements against the EUR. We pay special attention to the Brexit effects over the British constituents as it might represent a breakpoint in terms of trend. By analysing weekly returns of 55 index constituents in a 1-year period we identify a compelling pattern: around 89% of the stock s volatility in the Eurozone is related to its own dynamics (volatility not explained by external factors) and 11% to external factors (3.26% local equity market index, 3.27% to the real estate index, 4.1% to the real estate sector index). In the case of non-eurozone constituents, their own dynamics represent 81% of volatility and 19% can be attributed to the external sources, including the 3.86% associated to the local currency and 7.23% to the equity market. II. Indices & stocks: Volatility evolution and key facts This section provides an overview of the trends in total return indices and volatilities at the continental and country level. Historical volatility corresponds to the 52-weeks annualised standard deviation. II.I DEVELOPED EUROPE INDEX: RETURNS AND VOLATILITY At the continental level, the FTSE EPRA/NAREIT Developed Europe Index showed a consistent growth since 29 reaching a 5y total return of 15.34% in December 216. In terms of return volatility, three main patterns can be observed during the last ten years: 1) a significant increase during the Global Financial Crisis (GFC) with a maximum of 48% in September 29; 2) a new peak following the 211 European crisis, reaching a local maximum of 25% in July 212; and 3) a more stable stage of levels below 2% over the last four years. The average volatility for the entire period (27-216) is 2.9% and 16.1% for the last five years. FTSE EPRA/NAREIT Developed Europe - Total return index FTSE EPRA/NAREIT Developed Europe - Return volatility % 5% 4% 3% 2% 1% % Jan-7 Jan-8 Jan-9 Jan-1 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Exhibit 1. Evolution of the EPRA Developed Europe Index over time, Returns and Volatility.

4 European Listed Real Estate vs European Property Market Correlation ( ):.84 Exhibit 2. Total Return: FTSE EPRA/NAREIT Developed Europe Index (m/m) vs. IPD Pan-Europe Index (y/y). Source: Bloomberg. On the other hand, it is important to point out that the property market is considered as the main driver of the listed real estate market. From Exhibit 2 it results clear the strong effect that the European property market (IPD Pan-Europe) has over the total return dynamics of the FTSE EPRA/NAREIT Developed Europe Index. Indeed, the correlation between the two indices over the period reached.84. However, listed real estate can also be more volatile as it is at the same time influenced by the property market and by some additional factors that determine the dynamics of both local and continental financial markets. II.II EXTERNAL FACTORS: EQUITY MARKETS AND CURRENCIES Comparing the return volatility of the listed European real estate to the non-eurozone currencies: CHF, GBP, NOK, and SEK, Exhibit 3 shows that the four currencies and the FTSE EPRA/NAREIT Developed Europe Index displayed similar reactions to the outburst of the GFC, increasing from August 28 to November 29, to then revert to levels superior but close to pre-crisis levels. After June 211, the patterns tend to diverge. From September 214 onwards, an ascendant trend is noticeable in the five return volatility series that is, the index and the four currencies. These observations indicate a potential sensitivity of the index to currency movements through its non-eurozone constituents.

5 Volatility Volatility Volatility RPRA and Currencies - Volatility 6% 5% 4% 3% 2% 1% % RPRA NOK SEK CHF GBP Exhibit 3. Historical volatilities: FTSE EPRA/NAREIT Dev. Europe index and non-eurozone currencies vs Euro. Finally, as depicted in Exhibit 4, the return volatility of non-eurozone (left) and Eurozone (right) national equity markets display similar patterns over time than the FTSE EPRA/NAREIT Developed Europe Index, however in most of the cases the equity markets show a slightly higher volatility than the listed real estate, especially in the Eurozone markets. These observations once again confirm the expectations that national equity markets do influence the constituents volatility. RPRA and Non-Eurozone equity markets - Volatility.6 RPRA and Eurozone equity markets - Volatility OSEAX SAX SPI ASX RPRA ATX BELSTK HEX PAX HDAX ITLMS AMX RPRA Exhibit 4. Evolution of volatilities over time: FTSE EPRA/NAREIT Dev. Europe Index vs. European equity markets. II.III CORRELATION ANALYSI S Correlations offer an additional layer of exploratory analysis. Exhibit 5 shows the correlation for the non- Eurozone constituents volatility against the benchmark s volatility (horizontal axis) and against volatility of the national currency (vertical axis). In most of cases, constituents volatility seems to be more closely related to the listed real estate index than the currency. Despite a concentration of points in the upper right corner, implying that high correlation with the index volatility often goes hand in hand with high correlations with the currencies

6 Correlations with currency volatilities, there appears to differences between nationalities. For instance, while they display high correlations with the index volatility, the return volatility of Swiss constituents (yellow) display low correlations with the Swiss Francs return volatility, while British constituents (blue) tend to concentrate in the upper right corner where correlations are high in terms of both British pound volatility and FTSE EPRA/NAREIT Index volatility. 1.1 Ten-year correlations between non-eurozone constituents' volatilities against currency return volatilities and RPRA index volatility Correlation with RPRA index Exhibit 5. Scatterplot: correlations on constituents volatility against the FTSE EPRA/NAREIT Developed Europe Index and its national currency. Note: Yellow - Swiss companies, Green - Swedish companies, Red - Norwegian companies, and Blue - British companies. Exhibit 6 plots the 52-weeks rolling correlations between volatilities of the FTSE EPRA/NAREIT Developed Europe Index (RPRA) constituents and their respective national equity market (non-eurozone, left and Eurozone, right). All correlations Exhibit similar patterns between November 28 and November 211, usually fluctuating between.5 and.93. However, the situation appears quite different outside the abovementioned timeframe. As is suggested at the very right of both graphs in Exhibit 6, different equity markets displayed different correlations with the RPRA index in terms of volatilities Correlation between volatilities of RPRA and Equity Markets - Non- Eurozone Correlation between volatilities of RPRA and Equity Markets - Eurozone Corr(RPRA, OSEAX) Corr(RPRA, SPI) Corr(RPRA, SAX) Corr(RPRA, ASX) Corr(RPRA, ATX) Corr(RPRA, HEX) Corr(RPRA, HDAX) Corr(RPRA, BELSTK) Corr(RPRA, PAX) Corr(RPRA, ITLMS) Exhibit 6. Correlations between national equity markets return volatilities and the return volatilities of the RPRA Index.

7 After November 211, the correlations also significantly diverged from their high levels, in many cases even turning negative in 214. However, in most of the cases correlations decreased in a sudden manner in 214 to then revert to positive levels even in 215 or 216, which may suggest and temporal shock affecting the entire volatility-correlation dynamics of the listed real estate companies and equity markets rather than a structural change. However, knowing that similar patterns were observed during the GFC (28-29) and the European debt crisis ( ), it might be possible to identify different dynamics during crisis and non-crisis periods. Nevertheless, it results clear that equity markets have a significant influence over the listed real estate volatility. Finally, we check the dynamic correlations between the return volatilities of the RPRA index and the return volatilities of the non-eurozone currencies. Generally, correlations are positive as a positive currency return (appreciation against the euro) represents a positive impact over the FTSE EPRA/NAREIT Developed Europe Index, which is based in EUR. Exhibit 7 reveals a dynamic in the evolution of correlations. As with the equity markets, correlations between the RPRA index and the non-eurozone currencies appear to move together for most of the considered period, except for the period surrounding the 211 European crisis, further confirming time-dependent patterns and suggesting a less significant effect of currencies volatility over the index s volatility. Exhibit 7. Correlations between non-eurozone currencies volatilities and the return volatilities of the RPRA Index Considering the results from the exploratory analysis performed in this Section, we observe from the evolution over time of their return volatilities that the FTSE EPRA/NAREIT Developed Europe Index, the national equity markets and the non-eurozone currencies have been moving closely together over most of the considered period, more so in some instances than others. This suggests that index constituents are likely to be sensitive to movements in their national equity markets and their respective currencies, when applicable. However, this sensitivity appears to differ across countries and sectors, as suggested in Exhibits 5, 6 and 7, which makes relevant to analyse and decompose the volatility of each constituent. In the same way, all 55 index constituents stock volatility display a strong correlation with that of the index. These findings thus confirm the influence that currencies, benchmarks and equity markets have on the FTSE EPRA/NAREIT Developed Europe Index constituents volatility, as well as suggest complex and time-dependent dynamics.

8 Frequency Frequency III. Data and methodology As common with most return series, real estate stocks and indices in Europe show complex dynamics with timedependent patterns. Leptokurtic distributions, clustering stamps and extreme values are recurring characteristics of most index constituents returns, thus complicating volatility analysis and calling for the use of complex tools. As an example, Exhibit 8 compares the total returns distributions of British Land (BLND) and the FTSE EPRA/NAREIT Developed Europe Index (RPRA). FTSE EPRA/NAREIT Developed Europe Index vs British Land BLND RPRA BLND RPRA Exhibit 8. FTSE EPRA/NAREIT Developed Europe Index vs British Land - Weekly returns: Jan/6 to Dec/16 Our dataset is composed of weekly returns during the period January 26 December 216 for 55 companies that have consistently been constituents of the FTSE EPRA/NAREIT Developed Europe Index over the considered period. Aware of the bi-directional influence between, on the one hand, individual stocks and, on the other hand, both equity market and listed real estate indices as well as currencies, we opted for a standard Autoregressive Vector Regression Model (VAR) with 12 lags for each constituent to capture the time-dependent dynamics and identify long-term patterns. Five variables are considered: constituent returns, RPRA returns, RPRA-sector returns, equity market index returns and currency returns. Next, we apply a variance decomposition technique 1 and we aggregate the results 2 using the constituent weights in the FTSE EPRA/NAREIT Index series for the corresponding aggregation level: country, sector and Eurozone vs non-eurozone. The component labelled as own dynamics represents all the factors exclusively associated with the company s performance but also the constituent s volatility not explained by the four remaining variables. In addition, assuming the volatility to behave as a random variable with the purpose of recognising its long-term determinants, we also apply the variance decomposition to the historical volatility series and we aggregate the results in the same way as before. Finally, the Brexit effects are analysed by testing for break points in the volatility series using a standard ARIMA model over the period and comparing changes between the pre-brexit and post-brexit announcement dates. 1 Cholesky Variance Decomposition: Results correspond to decomposed variance for the first variable in the VAR (constituent returns). 2 Convergence results, usually very close to those observed after 52 weeks. See Appendix 1.

9 IV. Volatility decomposition and long-term determinants IV.I VOLATILITY DECOMPOSITION : EUROZONE VS NON-EUROZONE We begin by comparing the decomposition of return volatilities and aggregating at the Eurozone and non- Eurozone level. Exhibit 9 reveals that for Eurozone constituents, around 3.26% of the stock s volatility can be attributed to the local equity market index, 3.27% to the real estate index, 4.1% to the real estate sector index, whereas the remaining 89.37% can be attributed to its own dynamics. In the case of non-eurozone index constituents, on average, around 7.23% of their stock s volatility can be attributed to the local equity market index, 3.43% to the real estate index, 4.44% to the real estate sector index and 3.86% to currency movements, while the remaining 81.5% to its own dynamics. In other words, investors might consider the effects of external factors when managing market risk for real estate stocks in Europe. In fact, around 1.63% of the stock s volatility of Eurozone constituents and 18.95% in case non-eurozone constituents could be hedged using financial instruments associated to those factors (i.e. options, ETFs, futures). Exhibit 9. Variance decomposition of returns. Note: Weighted average using starting period Index weights. 3 IV.II VOLATILITY DECOMPOSITION: DIFFERENCES ACROSS COUNTRIES Considering the results presented in Section 4.1 that suggest differences between Eurozone and non-eurozone constituents variance decomposition, the analysis is further disaggregated across countries. Various interesting findings can be inferred from Exhibit 1. Firstly, the Eurozone countries provide similar results: around 9.8% (Netherlands) and 12.9% (Austria) of the stock s volatility is attributable to external sources, such as the local equity market and the real estate indices. There is however an exception: Germany. For German constituents, 15.8% of their stocks volatility seems to be related to external sources, with 8.98% attributable to the real estate sector index which is mainly dominated by residential real estate companies in this case. When it comes to the non-eurozone constituents, external factors are responsible for between 14.7% (Norway) and 19.3% (UK) of their stock s volatility. The additional volatility arising from the currency accounts for part of this increment, and the equity market also appears to take on a greater portion of the volatility for Swedish and British constituents than for other constituents. In addition, it is also interesting to note that in most of the cases the effect of the continental listed real estate market (RPRA) is lower than the domestic equity market, in particular for the non-eurozone constituents. 3 Using the end of period weights do not significantly alter the results.

10 Constituents Volatility Decomposition Own dynamics Equity Market Currency RPRA Sector Exhibit 1. Variance decomposition of returns: Aggregated by country. IV.III DIFFERENCES ACROSS REAL ESTATE SECTORS Additional insights also arise from aggregating the variance decomposition results at the sector level. The analysis is once more performed for Eurozone and non-eurozone index constituents separately. In the case of Eurozone (Exhibit 11), the portion of volatility attributable to external sources vary significantly, from 4.17% for retail-focused constituents to 2.4% for residential-focused constituents. Eurozone Constituents Volatility Decomposition External Factors Equity Market RPRA Sector Exhibit 11. Variance decomposition of returns (Eurozone constituents): Aggregated by sector. Note: Weighted averages, excluding constituents that changed sectors

11 The impact of the local equity market differs across sectors and has the biggest influence on the office-focused constituents (5.96%), whereas it has the smallest influence on retail-focused constituents (1.65%). The latter constituents are also the least affected by the overall listed real estate market (.86%), whereas around 3.6% of diversified, office and industrial constituents volatility is attributable to the overall listed real estate market. When it comes to the importance of the sector itself, it appears to explain a staggering 14.79% of the volatility of residential index constituents returns, while it ranges from 1.66% to 3.26% for the rest of property specialisations. When considering the non-eurozone index constituents (Exhibit 12), external sources account from 12.71% (Healthcare) to 25.8% (Industrial) of the fraction of their stock s return volatility, which is significantly higher than their Eurozone counterparts. Out of these percentages, currency effects are responsible for 1.13% to 5.8%, and the residential-focused constituents are once again the most impacted by sector-specific dynamics, but the fraction of return volatility attributable to the latter sector-specific dynamics (7.52%) is around half of that reported for Eurozone constituents (14.79%). Note that the local equity markets also assume a greater role in non-eurozone constituents case than in their Eurozone counterparts, ranging from 2.86% (Healthcare) to 9.14% (Industrial), compared to the 1.65%-5.96% range observed in Eurozone constituents case. Non-Eurozone Constituents Volatility Decomposition External Factors Equity Market Currency RPRA Sector Exhibit 12. Variance decomposition of returns (Non-Eurozone constituents): Aggregated by sector. Note: Weighted averages, exclude constituents that changed sectors. IV.III VOLATILITY AND PROPERTIES: THE CASE OF THE UK Although our purpose is to decompose the constituents returns volatility through the mentioned factors, it is interesting to understand the role that properties returns (IPD Index) play in these dynamics that was mainly captured under the own dynamics component in the previous analysis. However, properties returns show a different timing (monthly, quarterly, yearly) than financial markets (weekly, daily, intraday), then we use two different time frameworks to analyse the FTSE EPRA/NAREIT UK Index (RPUK) as one of the most representative Real Estate markets in Europe.

12 RPUK ASX GBP RPRA DIVERSIFIED IPD-UK Exhibit 13. Normalized indices: EPRA UK (RPUK), UK Properties (IPD-UK) and all other factors. We start by applying the volatility decomposition to the total return indices in levels on a quarterly basis with the purpose of identifying the long-term changing aspects (Panel a, Exhibit 14). In this case we see a clear effect of the properties returns, almost 5% of the long term EPRA UK Index volatility can be associated with this factor and 88.37% with both own dynamics plus properties, which is in line with the results showed in Section 4.1 where the aggregated decomposition from UK constituents showed 81% of the volatility to be associated with the own dynamics factor. In the same way, when using the quarterly returns for all the variables, we can identify the mid-term pattern: the equity market plays a significant role generating 55.73% of the RPUK volatility and properties returns seem to have less significant impact (6.2%). a) Volatility Decomposition for the UK Including Properties Returns Levels: Q/Q b) Volatility Decomposition for the UK Including Properties Returns Returns: Q/Q Sector (1.85%) Properties (47.2%) Own Dynamics (41.35%) RPRA (3.74%) Currency (7.69%) Properties (6.2%) Own Dynamics (24.79%) Equity Market (55.73%) Sector (.97%) RPRA (4.84%) Currency (1.9%) Equity Market (3.92%) RPRA (.47%) Currency (6.62%) Equity Market (4.2%) c) Volatility Decomposition for the UK Including Properties Returns Returns: M/M Sector (.57%) Properties (1.34%) Own Dynamics (86.79%) Exhibit 14. Variance decomposition of EPRA UK Index including properties.

13 Finally, we apply the variance decomposition to the monthly returns of all the variables to get as close as possible to those results showed in Section 4.1 and 4.2 (weekly returns). Once again, the properties returns have a less significant effect (1.34%) than the long-term figure (47.2%) and the own dynamics component is the most relevant (86.79%). This could be associated with the fact that many of the market-participants buy and sell RE stocks daily or weekly without observing specific fundamental changes in these factors, including properties returns. In brief, around 5% the own dynamics factor that is mentioned through all the Sections of this report can be associated with the properties returns on a long-term perspective, however from a mid-term and shortterm viewpoint, the properties returns are less relevant and some other factors like the equity market trend, the currency volatility and trading activity appear to be the main sources of volatility. IV.IV LONG-TERM DETERMINANTS In this Section, we apply the same methodology than Sections 4.1 and 4.2, but instead of using returns as inputs we use the 52-weeks standard deviation of total returns and then we check for convergence in the variance decomposition factors, this allows us to identity the main sources of changes in the long-term volatility dynamics. Again, we just aggregate the results using each constituent s weight. Long-term Volatility Determinants United Kingdom Switzerland Sweden Norway Netherlands Italy Germany France Finland Belgium Austria Own Dynamics Equity Market Currency RPRA Sector Exhibit 15. Long-term variance decomposition of constituents volatility. Note: Weighted averages, excl. constituents that changed sectors. Exhibit 13 shows a noteworthy pattern: considering only the Eurozone constituents, in average 41.6% of changes in the constituents volatility has been related to changes in the volatility of some of the external factors (15.45% the local equity market, 16.73% the real estate sector performance and 8.88% the RPRA Index) and 58.94% can be associated with the company own dynamics (properties, corporate performance and others). For the non-eurozone constituents, this percentage rises to 55.91% (29.21% the local equity market, 11.86% the local currency, 6.88%the real estate sector performance and 7.87% the RPRA Index). Once again, the equity

14 markets seem to be the most relevant external factor affecting the volatility of the listed real estate in both Eurozone and Non-Eurozone countries. There are some interesting cases like Germany and Norway, the first since external factors are responsible only for 19.3% of the changes in the constituents volatility and the second where the local equity market s volatility shows the highest impact over the constituent s long-term volatility dynamics. IV.V BREXIT Last of all, we turn our analysis over the effects that the Brexit process has created on the FTSE EPRA/NAREIT UK Index constituents volatility. Using the same sample of companies than before, we focus on the period December 212 to December 216 which was previously identified as a stable stage in comparison with the peaks observed in 29 and 212. For this purpose, we consider September 215 as the best candidate to be a breakpoint since in that month most of the constituents showed a change in their returns trend once the British parliament cleared the way for a Brexit referendum and markets started discounting a higher risk in the UK real estate industry. However, that change of trend in the constituent s returns not necessary implies a breakpoint in terms of constituent s volatility. We tested 4 all the 2 companies included in the initial sample and we identified statistically significant breakpoints only for four of them: British Land (BLND), Great Portland Estates (GPOR), Land Securities (LAND) and SEGRO (SGRO)..9 Brexit effects: break points on historical volatility.8.7 BLND LAND GPOR SGRO Exhibit 16. Constituents who showed a significant breakpoint in the period Jan/12 Dec/16. Three of these four constituents show two relevant common facts: 1) they are classified as Diversified in the sectorial classification of the FTSE EPRA/NAREIT Developed Europe Index, 2) they have a substantial exposure to the property market in London. In all the three cases, either the coefficient associated to the equity market or the one related to the sector index switched from negative to positive or increased its magnitude after September, 215 then showing a negative Brexit-effect (higher volatility). However, at the same time the coefficients associated to the RPRA index turned negative for BLDN and GPOR and remained almost unchanged for LAND, implying that this negative effect was not a strong as initially thought. Finally, the fourth company (SGRO) shows a different kind of breakpoint (intercept change) suggesting only a temporal change that is not necessary associated to the Brexit effects. Appendix 2 shows a short summary of these results. 4 Tests were run using the Chow stability test for potential individual breakpoints.

15 V. Conclusions Overall, the findings presented in this report indicate that external factors play a significant role on the volatility dynamics of the FTSE EPRA/NAREIT Developed Europe Index constituents. Although more than 8% of each constituent s volatility is related to its own dynamics (including properties returns), external variables such as the listed real estate market, the national currency and the domestic equity markets should be considered in the portfolio and risk management process. In fact, in most of cases the equity market is the largest external contributor to the constituents volatility. For Eurozone constituents, 11% of the 52-weeks return volatility can be attributed to the influence of such external factors like listed real estate market and local equity market. This influence is even higher (19%) in the case of the non-eurozone constituents, partially due the volatility of the national currency but also as a result of a higher influence of the domestic equity market, particularly in Sweden and the UK. Results across countries in the Eurozone are mostly homogenous with around 4.3% of the constituent s volatility attributed to the listed real estate index and 3.3% to the domestic equity market. The exception is Germany with a significant influence of the real estate sector index, mainly residential. For the non-eurozone countries results differ: the domestic equity market represents more than 7% of the constituents volatility in Sweden and the UK while the same proportion is below 5% for constituents in Switzerland and Norway. Alternatively, aggregated results at the sector level show that constituents volatility in highly procyclical sectors as Industrial or Residential are more dependent on the influence of the domestic equity market. Constituents of the FTSE EPRA/NAREIT Residential Index in the Eurozone can be considered as the exception since almost 15% of their volatility can be attributed to the sector s dynamic. When including total returns from properties, results do not seem to change, although there are some significant differences between time horizons. In the long-term, volatility coming from properties returns is responsible for almost 47% of the EPRA UK Index, however in the mid-term and short-term this contribution is much lower (24.8% and 1.3% respectively) since some other factors like the equity market trend and frequent trading activities strongly influence the constituents dynamics. Finally, there is not robust evidence of a strong structural change created by the Brexit. Only four of the twenty UK constituents included in the sample show a significant breakpoint which seems to be closely related to the company s exposure to the London s property market.

16 VI. Appendix VI.I CONVERGENCE ON VARIANCE DECOMPOSITION: SELECTED GRAPHS British Land: Returns British Land: Volatility 1 Variance Decomposition of D(BLND,1) 1 Variance Decomposition of BLND D(BLND,1) D(ASX,1) D(GBP,1) D(RPRA,1) D(DIVERSIFIED,1) BLND ASX GBP RPRA DIVERSIFIED Unibail-Rodamco: Returns Unibail-Rodamco: Volatility 1 Variance Decomposition of D(UL,1) 1 Variance Decomposition of UL D(UL,1) D(PAX,1) D(RPRA,1) D(DIVERSIFIED_RETAIL_UL_,1) UL PAX RPRA DIVERSIFIED_RETAIL_UL_ Exhibit 17. Convergence on Variance decomposition. Returns (Left), Volatility (Right). VI.II BREXIT-BREAKPOINT RESULTS Volatility Regression BLND GPOR LAND SGRO R2.56 Brexit.32 Brexit.543 Brexit.7 Brexit Adj. R2.547 Coefficient.316 Coefficient.524 Coefficient.693 Coefficient Sector Diversified Diversified Diversified Industrial Intercept....1 Equity Market * * Currency * RPRA * Sector * Non-statiscally significant Exhibit 18. Coefficients pre & post Brexit breakpoint. All the variables in first difference.