The disappearing carry trade

A difficult investment climate has caused a flight to safety from currency carry. Iain Morse assesses what other types of style approach are available for investors

Whatever happened to the carry trade? For a good 18 months until last summer, going long on high interest rate currencies and short on low interest rate currencies seemed an infallible formula for making money from currency. "Carry was consistently successful," recalls Torquil Wheatley, director global risk strategy FX at Deutsche Bank in London.

"Other styles such as mean reversion or fundamental looked less successful," says Wheatley. Many new investors wanted to participate in this new asset class; over the same period many pension funds invested in currency for the first time. Carry became a huge driving force in currency, with the resulting flows of money affecting currency values.

"Many of the non-carry traders capitulated, and turned to carry. By the start of last year the sheer success of carry and preparedness of investors to follow this success were sweeping the market. Then it all went pear shaped," continues Wheatley.

"July and August saw massive de-leveraging and a flight to safety," recalls Julian Samways, partner at Harmonic Capital Partners. "Some traders were caught in extended positions and had difficulty getting out of these." This was true of dedicated currency funds, but even more so of the broader hedge fund community which had come to see currency as an easy, accessible asset class. With the benefit of hindsight this was hubris on the part of some managers and some of their investors have paid a very, very high price for it.

"The key is not to lose money or at least to consistently minimise losses," assesses Diane Miller, head of currency research at Mercer. "This is the biggest differentiator between currency managers." The implication of this is clear. Currency markets display inefficiencies and skilful managers can extract returns from these. Whether any one style does this more consistently than the others is open to question. Indeed, there are several styles of currency management which can be run on a free-standing basis or blended into risk adjusted portfolios. "The bottom line from an investor's point of view is not the fine detail of their methodology but the skill with which they apply internal risk controls to these, and avoid losing gains when they are wrong," adds Miller.

There are good reasons for putting this emphasis on risk control, not least the difficulty of clearly distinguishing between currency management styles. "Managers migrated between styles, but the events of June and July 2007 were a lesson for traders who did not unwind their positions quickly enough," adds Wheatley.

Category distinctions in currency management are often inexact and managers not always keen to be explicit about their style. Roughly speaking, most analysis allows a distinction between value and return forecasting and between trend following and non-directional strategies.

Value forecasting uses fundamental factors, external to the currency market, to take currency positions. The emphasis here is on using economic analysis to determine whether one currency is mis-valued relative to another. Return forecasting is closely related but places more emphasis on the direction of future returns rather than on current currency valuations.

Both can be characterised as fundamental styles. Next comes trend following. This eschews fundamental analysis and focuses instead on mining past data to identify and exploit valuation trends. As long as a trend endures, this approach can yield stellar returns. The problem comes when the relative value of a currency pair reaches an inflexion point, and reverses the trend.

Arguably, trend following may be less successful at predicting these inflection points than more fundamental styles. Last but not least, non-directional styles seek to exploit volatility rather than longer term trends or valuations. Volatility levels may vary but some volatility is always present in the currency markets. This can be exploited by trading currency options; this is often done with a bias to being short volatility.

A further distinction between ‘judgemental' and ‘model-driven' or ‘technical' can be applied to any manager using one or more of these styles. A judgemental house is typically characterised as one permitting human intervention to override any buy or sell decision generated by any relevant model. A model-driven approach implies the reverse. This begs the question that any model is essentially a series of pre-programmed buy or sell decisions and will be subject to revision.

The reality is that most houses will employ one or more models for at least indicative purposes, but will also allow ‘manual override' if the model starts to generate excess losses. It is the definition of excess loss that can vary between managers using generically very similar styles. It is also clear that some styles are more suited to judgemental management and some to model-driven. For instance, fundamentals sit more comfortably with judgemental management and volatility with a model-driven management. But there are no fixed lines here; some models are fundamental, some judgements based on reading of intraday market sentiments.

None of these approaches is decisively proven as superior to the others. This is surprising given the size and liquidity of the currency markets, which have a daily turnover in excess of $2trn (€1.3tn), compared to around $500bn for US government bonds and $70bn on the New York Stock Exchange. In theory, this should be a truly efficient market. The consensus is now that it is not.

This raises some interesting questions about participants in the currency market. Traditionally these have been divided in two - forced participants and profit seekers. Forced participants must take currency risks as ancillary to other commercial operations. Profit seekers understand this and try to benefit from the market inefficiency created by the forced participants. Dealers now comprise a third group of participants. Their role is as intermediaries between forced and profit-seeking participants, and they provide an important source of short-term liquidity in this market.

Estimates vary, but there is broad agreement that profit seekers comprise no more than 50% of the market. This should leave them ample opportunity for profit, but it leaves open the question of whether for profit activity, particularly in emerging currencies, exceeds this 50% upper limit. "No one is really sure of what drives the currency markets." says Wheatley. "You cannot time the market and style hopping - choosing a style for the current stage in the cycle - doesn't work either."

The fact remains that a growing number of currency managers, particularly those which are part of global banks and asset managers, now place an emphasis on analysing currency flows and trying to second guess what their rivals are doing. Behavioural factors are seen as quite important by many for profit participants. "The impact of trades is fairly short term but it can be exploited," says Andy Bound, manager and head of fundamental currency at Goldman Sachs Asset Management.

None of this is very comforting for pension fund managers and boards. Those that have lost money from the carry trade since last summer may now think of leaving this asset class. But help is at hand.

A number of currency style indices are now available. Mercer uses a suite of four devised by James Binny, director of FX analytics and risk advisory at ABN AMRO. "These are not products and we do not allow funds to replicate their performance. They lack risk control or stop loss, but can be used as benchmarks," Binny explains. The first is a measure of the return from value forecasting using purchasing power parity; the second trend based, the third yield based, and the fourth is based on short volatility.

‘Value' is based on purchasing power parity, ‘trend' on a simple moving average strategy. ‘Yield' buys the higher yielding currencies and sells lower yielding ones. ‘Short volatility' sells an at-the-money-forward straddle for each currency at the start of each month if the currency's implied volatility is above its one-year average. "These could be described as naïve benchmarks, but they say something about the distribution of returns by each strategy," adds Binny.

Over the 12 months to the end of March 2008, the value index returned 1.2%, trend 1.3%, yield -3.5% and short volatility -6.1%. These returns do not include any underlying cash return, and in actively managed currency mandates the benchmark would normally be set at LIBOR plus. "If you cut up the data over different periods then the relative performance of these strategies can look different," adds Binny. For instance, in 2004, value returned 0.21%, trend -0.31%, yield 1.38% and short volatility no less than 18.66%. During March the same respective returns were 1.9%, 0.3%, -0.3% and -2.3%.

This return disparity helps to explain the growing popularity of currency fund of fund and fund platform products. "Judgemental management is better at identifying and responding to sudden change," argues Bound. "It has the big advantage of allowing a manager to step back from the market at certain times decide if a ‘regime change' is taking place."

Consultants and clients may feel more comfortable with model-driven styles of currency management but the in the long run judgemental styles may prove more optimal.