The new European currency, due to start trading on January 4 1999, will have no history. This trite observation has tremendous consequences for investors and traders, risk managers and regulators.
Why? Because all of them have come to rely, to a greater or lesser extent, on time-series data and historical volatility as a guide to pricing the risk they are taking and the likelihood of price movements in the future. Suddenly, history disappears.
For chartists, who like to follow trends, there will be no trend. For data providers, who like to supply their customers with nice graphs, the database has to start from scratch. For option traders, who like to check prices against historical volatility, there is none. For bond traders and investors, the performance parameters of many bonds will change. For risk managers who want to calculate maximum possible losses to a 90% confidence level, the confidence floor has temporarily disappeared. For regulators and supervisors there are no data against which to back-test value-at-risk numbers.
This is a potential nightmare not just in Euroland, but anywhere where institutions have big euro positions. The good news is that the nightmare will be temporary: the uncertainty will decay daily as tick data for the euro accumulate. But the Basle committee on banking supervision recommends at least one year of data as a basis for value-at-risk calculations, and some supervisors would like three years for greater comfort. So, for up to three years there will be a fudge.
Market participants are beginning to address this problem – just one of the many mysteries they have to come to terms with before 1999. In the front line are the data providers. They have already seen demand from customers for some kind of synthetic euro history. A Bloomberg user group met on September 12 to swap ideas on whether such a history would be useful and how it could be created. Reuters, Dow Jones Markets, and Bridge Information Systems are all consulting their customers and offering sample time series.
Another notable data provider is JP Morgan, whose RiskMetrics database relies heavily on historical data. Competitors are saying, perhaps too hastily, that RiskMetrics will “blow up” with the launch of the euro. Its function is to provide correlation data between instruments in major currencies: since the euro is a new currency, correlation data in relation to the European currencies must start again from January 4 1999. “RiskMetrics’ daily volatility estimates basically need 74 days worth of data,” says JP Morgan’s own literature and warns: “For the first few months of 1999 risk managers will need to identify proxies to be used in the risk measurement process.” Customers expect JP Morgan to offer something soon. Says Guy Coughlan global head of risk management products at JP Morgan in London: “Providing choice is the objective, just as we have different choices of co-variance. We don’t want to force the market into a view.”
Allowing the market to develop a consensus, but not necessarily one standard, also seems to be the regulators’ goal. “It’s up to the market,” says Paul Mercier, senior market analyst at the European Monetary Institute, forerunner of the European Central Bank. “But it’s clear we’ll try to establish our own statistics.” Says another European central banker: “We haven’t made any analysis of this. We should ask the banks in which way they are dealing with databases.” The Bank of England believes its role is to stimulate debate and heighten market awareness, without itself prescribing a view on synthetic price history.
Passing the parcel
“It seems to be a case of pass the parcel,” says one information vendor. “Some say: ‘We think it’s about time the vendors provided a package’. We reply that we only publish others’ information. But now we find we have to take a stance.”
The regulatory aspect is concentrating the minds of the major trading houses: from January 1998 they will be submitting their value-at-risk calculations for approval by supervisors, to determine the regulatory capital they must hold. If for any reason the supervisors don’t trust a particular risk model or its application – and that could include its estimate of future euro volatility – a house may be forced to use a standard calculation for value at risk to set its regulatory capital requirements. Practitioners suggest the standard calculation would cost a big trading house roughly four times the regulatory capital that use of an internal model would indicate. So, whether they do it in public or private, the big houses are interested in producing a surrogate history of euro volatility good enough to satisfy their supervisors.
Reuters has published some sample pages of synthetic euro interest rates, using Ecu history, but pointing out that other price histories could be used. At Dow Jones Markets, says one source there, “we’ve created pages to show the market we’re thinking about the issue”.
Purists view any attempt at producing a synthetic history as a fudge. “Any solution will be based purely on assumptions,” warns Chris Golden, director of European monetary union (Emu) research at Nomura International. But he adds that a history “might be a reasonable language to use”, just as quoting a bond’s yield to maturity makes assumptions about reinvestment rates, but is nevertheless a useful standard.
Three classes of data would be needed: the euro exchange rate against major currencies; euro interest rates; and synthetic bond-price history.
Likely surrogates
For the exchange rate there are three main candidates to act as a surrogate: the Deutschmark; the Ecu; and a weighted basket comprising the currencies that will join the euro. The choice between these candidates depends on how strong you think the euro will be and how good you think the European Central Bank (ECB) will be at keeping the currency stable. A strong euro might behave much like the Deutschmark, so a Deutschmark history might do. But if the euro is weak, then it’s more likely to be replicated by Ecu history, or by a basket that includes the volatility of the Italian lira and the French franc. However, none of these can behave exactly like the euro, since the euro is a new currency, and the world – especially with regard to international trade and investment patterns – has changed.
At preliminary brainstorming meetings a majority of information users have voted for the Deutschmark as surrogate. “But as you get into Emu you may want to change your mind,” suggests Matthew Elderfield, director of European policy at the International Swaps & Derivatives Association (Isda).
The Deutschmark model could give way to something more like the Ecu. Some risk models, such as JP Morgan’s RiskMetrics, give more weight to recent data. As JP Morgan’s literature argues: “The standard RiskMetrics approach which exponentially weights market data for the purpose of estimating volatility will prove superior in this environment to models which take longer to adjust.” So whatever surrogate is chosen should be designed to lose weight against more recent real euro data as they come in. “The Ecu and the Deutschmark will represent the two extremes,” suggests a London-based risk manager, “somewhere in the middle will be compromise.”
Some participants favour the Ecu as a surrogate, although it’s not a proper currency. The advantage of using the Ecu is that it will be exchanged 1:1 with the euro. But its past behaviour will give little indication of how the euro will behave. “The Ecu is an internally driven basket (driven by factors within the EU),” says Bob Blower, outgoing associate director of the British Bankers Association. “The Ecu is not a good proxy for externally driven factors.” Volatility experts warn than an average will artificially dampen volatility; but that is not true of exchange rates if they are correlated, like those between currencies in the European exchange rate mechanism (ERM) says Neil Record, chairman of Record Treasury Management. “With currencies there is no such concept as volatility,” he says. “It exists only in bilateral relationships with other currencies.”
Picking a trade-weighted or GDP-weighted basket of the future euro currencies – when they are picked and bilaterally fixed next May – is intellectually more attractive. But there are snags. Although bilateral rates between these currencies will be known, the external exchange rate of the euro – against the dollar and the yen for example – won’t be known until 1999. Nor will the exact relationship of the currencies joining the euro to the euro itself – because some currencies in the Ecu basket, such as sterling and maybe the lira, will be left out. As soon as these unknowns become known, on January 4 1999, and the exact composition of the synthetic euro basket can be decided, it will still require enormous computer power to back-calculate a weighted synthetic price history for this basket. Moreover, a synthetic euro which includes Italian lira volatilities for 1992, when the lira left the ERM, may show greater volatility than a euro could credibly have experienced at the time. Should this be fudged out of the model?
Trade secrets
One suggestion is to create a euro index, like the trade-weighted dollar index. But if a bank has gone to the trouble and expense of selecting and calculating such a basket and its history, it would probably want to use it for competitive advantage rather than share it with the entire market. “This is an area where people tend to keep their mouths shut,” says one data provider. Says another: “The top trading houses will welcome a standard – and go their own way. Just as they invent their own yield curve, they’ll have their detailed view on the history.”
But traders at major trading houses candidly admit they don’t trade options by looking at historical volatility, synthetic or otherwise. “Our options traders don’t care that much about historical volatility,” says one risk manager. “It’s a very seat-of-the-pants area. They tend to price off the forward values.”
Says a currency trader: “Historical data isn’t that important for trading. But if everyone else looks at it we have to look at it too. It’s a good sales pitch.”
Technical traders will need to measure their performance against some historical benchmark, especially if they are dollar-based, says one information vendor: “If you’re European, with a European customer base, a synthetic euro history doesn’t matter that much. But if you’re a US hedge fund with European positions, it does.”
The pre-announcing of bilateral rates next May for currencies joining the euro in January 1999 won’t reduce the problem, in fact it could double the frustration. “From the chartist’s point of view this will be a nightmare,” says Dick McDonald, vice-president for research at the Chicago Mercantile Exchange. “In May 1998 they’re changing the world in the way chartists look at it, and then in 1999 they’ll change the world again.”
Interest rates, especially defining a “risk-free” reinvestment rate, are a different challenge. Their synthetic history also depends on the power and credibility of the ECB. Looking at a basket of national interest rates is unlikely to help, especially at the short end of the yield curve. Today these short-term rates vary greatly, depending on the central bank’s view and national economic policy. “There’s a need for central banks to converge on short-term rates,” says a European bank strategist.
New bond models
Bond prices are a further conundrum. Euroland’s benchmark sovereign bonds will suddenly become defaultable, since national governments can no longer print money at will. “How do you incorporate default value into bond volatility?” asks one information vendor. “How do you define the risk-free rate?” If bonds are renominalized, and other conventions such as day counts are changed, that will also affect comparisons with their price and liquidity history.
Con Keating, chairman of the committee on methods and measures at the European Bond Commission, wrote in a note to the Bank of England working group on Emu: “Derivatives models will need modification to recognize the existence of default risk in the government security (a hedging instrument) and this will increase the cost of replication using more of the governments than previously whether long or short. Even then a credit concentration risk will exist which is not by convention present today.” Keating suggests that the closest to a default-free rate will be offered by the sovereign debt in euros of the UK, followed by Germany, France and Italy – in that order.
Information vendors will promote a choice of solutions for synthetic euro price histories. “JP Morgan and others don’t want a standardized approach because of the value of their databases,” says one euro expert rather cynically. But there is also a possible systemic danger in most of the market going for a single solution, which of its nature cannot be exact. Some European futures exchanges are likely to offer synthetic databases to help their members, and perhaps to fine-tune their own margining software. “We’ve been there already,” says Giuliano Gregorio, director of business development for Emu at Liffe, the London futures exchange. He’s referring to the synthetic history that Liffe has produced for the 8% notional Italian government bond (BTP) future. Last month Liffe reduced the coupon on its BTP contract from 12% to 8% to bring it more in line with converging European rates. “We received many requests especially from American customers for a historical time series [for the 8% bond future],” says Gregorio. The synthetic series (see chart) can be found on Liffe’s website (https://www.liffe.com). “We put it out, with disclaimers. You have to recalculate what the price would have been if the coupon had always been 8%. But you have to calculate a cash market price and a repo market price, which is impossible. So we ignored the repo rate and put in the Eurolira rate as the refinancing rate,” says Gregorio. “We produced something that never happened. But it’s very much appreciated, especially by technical traders.” The French futures exchange Matif has gone through the same exercise, having reduced the coupon on its 10-year notional OAT government bond from 10% to 5 1/2%.
Could Liffe produce an instant history for a notional government bond in euros? “If it’s the same old Bund [German government bond] contract there’s not so much of a problem,” says Gregorio. “If not, we’d have to make some assumptions, using the former Bund contract as a reference for part of the yield curve and perhaps the French for another part. Every time we launch a new product we have to make some assumptions,” he concludes.
One German banker compares the launch of the euro to the launch of Deutsche Telekom shares last year. “You have to find a proxy,” he says. “An equity issue always appears with no history.”
While there is still no consensus on these issues, various joint efforts are being made to focus opinion. Apart from the Bloomberg user group, the Bank of England has regular meetings with market participants, the last on September 29. And the foreign exchange sector’s Association Cambiste Internationale (ACI) is considering, among other proposals, commissioning a study on the value-at-risk parameters of the single currency. A decision on this is expected in mid-October and if it goes ahead some results would be seen by January, predicts ACI’s honorary president David Clark. The aim is to help regulators and practitioners to agree volatility parameters for the euro, so that an appropriate level of regulatory capital for euro positions can be set.
Wise people keep their minds open to the thought that no synthetic euro history will do – you can’t predict the future, especially one as revolutionary as this. “A whole industry has grown up trying to create problems that aren’t there,” complains one European strategist.
But for risk management it’s another milestone. “If Emu forces managers to reduce their sometimes blind faith in their market risk models,” says JP Morgan in its RiskMetrics literature, “risk management will have become a more mature practice.”
