Blending in turbulence
The euro’s introduction was clearly a seismic shock to the international financial system, with wide-ranging impacts across global markets. Not only have private financial institutions had to restructure their products and operations significantly. Politicians and senior public officials have also had to learn a new “official” language in discussing the new currency – witness the response in the currency markets to the resignation of German finance minister Oskar Lafontaine in April and the comments of Romani Prodi in June regarding Italy’s long-term ability to remain within the euro.
In this article we explore the impact the creation of the euro has had on a particularly important class of financial institution – the pension fund. We will consider some of the investment opportunities that this presents and the appropriate framework for considering their exploitation.
With regard to the euro, we can think of a pension fund as falling within one of three layers of the currency “onion” (see Figure 1). At the core of the onion are funds based in one of the 11 Emu member countries. They have generally felt the most significant impact, as their base currency is re-denominated, with the consequent administrative burdens imposed. The second layer comprises funds based within Europe but in countries outside Emu, principally Switzerland, the UK and Scandinavia. Their domestic economies are becoming increasingly impacted by the Euro-zone and the considerations of potential Emu membership at some stage. The third layer of the onion is of funds outside Europe for whom the euro is interesting, but probably more from a technical perspective than a structural change.
Pension funds at the core of this onion now have a unique investment opportunity. For the first time in recent history a significant proportion of the global investor community is able to invest in foreign securities without incurring any currency risk. This has had two significant impacts already. First, a dramatic rise in transnational investment has taken place within the euro-zone. Second, investor appetite has concentrated on euro-denominated issues, in turn attracting a new pool of issuers. Dramatic advantages accrue to a euro-based fund, which is now able to increase its pool of investable securities with the concomitant improvement in portfolio flexibility, liquidity and diversification. It will be interesting to see, in the medium term, if these benefits encourage further diversification into non-euro assets with even broader appeal, albeit with the necessary accompanying currency exposure.
Looking at the second layer, it is interesting to see that there is increasing enthusiasm here to join Emu after all. Denmark’s population was recently estimated in a poll by IFKA to be around 46% in favour, 34% against. The ruling Social Democratic party is set to debate holding a referendum in September 2000 at its annual party conference. Sweden’s Social Democrats will undertake a similar exercise when they meet next March. Although Norwegian polls indicate a small majority in favour, the country has yet to overturn the result of a 1994 referendum which rejected EU membership – a prerequisite for joining Emu. Furthermore there is considerable concern as to how the onshore economy, which is most likely to benefit from the euro, can be separated from the offshore oil industry, which is linked to the dollar. Switzerland appear most unlikely to join, less for economic reasons than the political, social and environmental authority that would have to be ceded to the EU. Finally, the largest non-euro European economy, that of the UK, remains predominantly “euro-sceptic”. Certainly the only political commitment is for consideration of the issue “early in the next parliament” – most likely 2001.
A pension fund’s approach to the euro will depend on where it lies within this onion. Nevertheless a common strand throughout will be the need to have an appropriate mechanism for working with this new currency, given its limited track record.
Early in the life of the euro – and indeed for some time prior to its formation – there was considerable debate about whether there is any relevant history that may be used to estimate the key risk variables that are necessary in guiding international investment: the volatility of the euro itself and its correlation with the other currencies and assets in the portfolio. Essentially the debate revolves around three options:
q The European currency unit (Ecu): not for the seriously minded! Its principal appeal is that it was an actual traded currency and therefore has a real history. Additionally, since it converted to the euro at a 1:1 exchange rate it provides the (spurious) implication of being equivalent to the euro. However, the fact that 20% of the Ecu was made up of non-Emu currencies rules it out for any realistic analysis.
q A synthetic basket: It would seem to make sense that the new zone will behave like the sum of its parts, with the more important parts playing a more significant role. One could therefore construct a “synthetic euro” by averaging the exchange rates of the euro-constituent currencies. The question then is what averaging approach to take: should the constituents be equally weighted, GDP-weighted1 or market capitalisation-weighted. However the euro economic zone is not merely the sum of its parts. The very fact of the single currency is just one of the structural changes that make it quite different from the former EEC, some of the others being the European System of Central Banks and the unified monetary policy. Indeed during 1997–98 we saw that as the 11 currencies approached convergence they behaved quite differently from earlier periods. So a synthetic basket is not entirely appropriate.
q Deutschmark: In the past year, observed volatilities of the Emu currencies from the US dollar had already converged to that of the Deutschmark. Given that the Emu currencies are almost perfectly correlated and have almost identical variances, it follows that one should be able to use any single Emu currency as a proxy for the euro. Though this is well suited to short-term forecasting (as used by RiskMetrics2, for example), is it safe to use the Deutschmark for longer horizons, which means having to go back further in time? A paper by Brooks, Scott-Quinn and Walmsley3 looked at the history of various baskets of European currencies since 1993, and found no statistically significant difference between any of the baskets (with the exception of the Ecu) and the Deutschmark.
Seven months into the euro, we are now able to draw some preliminary conclusions on the efficacy of these three models of the euro. In Figure 2, we show a score of how well each approach did in fact predict the actual risks of the euro evidenced so far. With its large UK component and sterling’s divergence from the euro, the Ecu provided a poor model for the euro. Although both of the other models perform reasonably, the dominance of the German economy appears to make the Deutschmark the most robust tool for modelling the euro.
Clearly, whichever approach is taken to calculating these risk estimates a new currency does bring with it increased caution about relying on historically determined values. At our firm we have addressed this problem by controlling our portfolio selection process for the likelihood of increased turbulence in the future. This is based on the thesis that by increasing the weighting given to unusual observations in the past we will gain superior insight into each portfolio’s likely performance in future similar circumstances. A recent paper in the Financial Analysts Journal4 explains the process in considerable detail however a summary intuition can be gleaned from Figure 3.
Here we see a scatter plot of two assets’ returns – each point representing one period’s returns. The shaded ellipse indicates where the “normal” observations occur – this is when the markets are quiescent and few events have a significant impact on the assets’ returns. Outside of this are the turbulent observations, where more extreme events occurred. It should be noted that these can occur either because one (or both) of the assets’ returns is abnormal – at the extremes of the green axis; or because the co-movement between assets is abnormal – at the extremes of the red axis.
While the individual “turbulent” observations can be readily observed from such a graphic, in reality we must resort to matrix algebra to locate them in a real-world portfolio – the reader is directed to the original article for a fuller explanation.
During turbulent times the most common problem for portfolio managers is that asset volatilities tend to rise and so do the correlations between the assets. Furthermore, for global portfolios, the same observation can be made for currencies and their relationship with the underlying assets. This is frequently observed to dramatically undermine the diversification effect that is the raison d’être of international investment. In the context of currency particularly, it should be recalled that the optimal currency hedge ratio is the sensitivity of the portfolio to movements in that currency’s value – its beta.5 This is a function of the correlation and relative volatilities:
Hi = ri,p ¥ sp__si (equation 1)
where Hi is the optimal hedge ratio, ri,p is the correlation between the currency and the portfolio and sp and si are the volatilities of the portfolio and currency respectively.
We can see therefore that a rise in the correlations implies higher hedge ratios than would otherwise be required.
Looking at a fund based in the second layer of the onion, we can illustrate these points by considering the effects of turbulence on a Swedish krona-based portfolio.
Firstly, let us take a look at a sample of the risk and return parameters for the whole period of analysis6 as well as quiescent and turbulent sub-periods (see Table 1). We can see, as anticipated, that the risk levels are significantly higher during the turbulent periods. This observation is consistent throughout the assets in this portfolio. The variation on the risk levels though is considerably greater for the international assets than the domestic ones. Furthermore, we are able to recombine the data by blending the quiescent and turbulent periods in such a way as to give somewhat greater emphasis to turbulence than was experienced in history, as indicated by the “Blended” column.
Taking a similar view of the assets’ returns, we see that these also are generally higher during turbulence (see Table 2). This, however, is a period-specific feature resulting from the devaluation of the Swedish krona raising the value of foreign investments and should not be expected more generally.
This process can then be used for a variety of stress-testing applications in asset allocation and portfolio selection deliberations. In the context of currency for example, the definition of a strategic hedge benchmark can be illuminated through the application of such regime-specific parameters to equation 1.
Table 3 then shows us that the appropriate strategic hedge might naively be computed at 70%. Certainly anything less than the quiescent 47% is likely to over-estimate the stability of the portfolio and a fully hedged strategy implies that the future is expected to be much more turbulent than history. The blended strategy’s hedge ratio of 78% should be a more credible yet conservative recommendation.
The euro has already had a significant influence on pension funds around the world. Initially these have been at the micro-level with accounting and asset-specific issues dominating. Increasing attention is now being given to the medium-term asset allocation opportunities that are becoming available. Currency management activities will undoubtedly be revised in the light of these. Further on into the future one can foresee more systematic results: greater labour mobility, regulatory harmonisation and competition between pension fund providers: We certainly live in interesting times!
Jeremy Armitage is vice president and head of currency management at State Street Associates in London